WO2002044990A2 - Apparatus and method for determining affinity data between a target and a ligand - Google Patents

Abstract

Apparatus for determining affinity data between a ligand and a target comprises calculator means (1) for receiving information representative of a target and of a ligand that might interact with the target at one of its receptor sites, and for determining an affinity value representative of the association energy between said target and the ligand on the basis of the received information, and comparator means (3) arranged to deliver data representative of an affinity level between the target and the ligand on the basis of the target/ligand affinity value and of a selected reference parameter.

Description

APPARATUS AND METHOD FOR DETERMINING AFFINITY DATA BETWEEN A TARGET AND A LIGAND

FIELD OF THE INVENTION

The present invention relates to an apparatus for modeling interactions between targets and ligands. The present invention also relates to a method of determining drug resistance, as well as a method for monitoring a patient who is becoming resistant to drug treatment.

BACKGROUND AND PRIOR ART

In numerous fields, it is important to be able to predict whether a ligand is suitable for interaction with a target. This is particularly important in the medical field, particularly when a disease progresses quickly. At present, it is not possible to quantify the association between a target and a ligand other than by comparison with other associations. Thus, with a virus (target) it is possible to say that it presents resistance to a first medicament (ligand) that is X times greater than the resistance it presents to a second medicament. In other words, the duration of association between the target and the first ligand is one-Xth the duration of the same target and the second ligand.

The person skilled in the art determines the resistance of a target/ligand entity or more generally of any other affinity parameter representative of a target/ligand association by in vivo or in vitro tests. Unfortunately, regardless of whether the test is performed in vivo or in vitro, it often happens that the particular environment in which the test is performed influences the results. Resistances obtained in this way can thus vary significantly from one test to another, and from one subject to another. That is why, in particular, it is generally possible in the specialized literature to find only statistical mean values for resistances, or for any other affinity parameter representative of a target/ligand association. Furthermore, certain in vitro tests for identifying target/ligand interactions require complex techniques to be implemented such as cell culture, sequencing, or phenotyping, which are expensive and can require several weeks of work, which can be incompatible with the mutation frequency of a virus (target), as is the case for the human immuno-deficiency virus type 1 (HIV-1 ), and more precisely for the protease binding site of HIV-1 and reverse transcriptase (RT). These targets can be inhibited (or neutralized) by inhibitors (ligands).

Medicines are known for inhibiting the HIV-1 protease binding site by blocking the late stage of virus maturation, and more precisely which act on replication within chronically infected cells. They are active on activated CD4 lymphocytes which produce virus, and also on cells that merely present antigens such as macrophages. Thus, HIV-1 protease cleaves precursor polypeptides produced by the "gag" and the "pol" genes that enable the structural and enzymatic proteins of the virion to be generated. In the presence of HIV-1 protease inhibitors, the virions which are produced are immature and thus incapable of infecting new cells.

Of HIV protease inhibitors, particular mention can be made of indinavir, ritonavir, saquinavir, amplenavir and nelfinavir. All of those protease inhibitors are medicines presenting a metabolism that operates by means of p450 cytochromes. That is why they are sensitive to drug interactions with other substances that use the same metabolic pathways, or that induce such drug interactions.

In general, a ligand (such as an inhibitor) interacts with a target (such as HIV-1 reverse transcriptase or protease) by means of at least one of its receptor sites. Such a receptor site can be defined by a continuous portion of the target or by a plurality of discontinuous portions of the target, e.g., when the portions are adjacent because of folding. For the protease binding site of HIV-1 , a receptor site is defined by a certain number of amino acids traditionally referred to as "residues". It is these residues which are implicated directly or indirectly in target/ligand interaction (or association). As mentioned above, one of the major drawbacks presented by the protease binding site of HIV-1 is its ability to mutate rapidly (it is certain residues which are actually subject to mutation). As a result any drug (such as an inhibitor) having the reputation of being effective with respect to a reference non-mutated protease binding site of HIV-1 can often be partially or even completely ineffective with reference to at least one of the mutations of the reference protease. In other words the target (in this case the mutated protease) has developed resistance to that ligand (in this case an inhibitor medicament). There are lists giving the known mutations of the protease binding site of HIV-1. One such list contains, for example, mutations referenced M46L, I54L, A71T, I54V, L90M, V82S, I84T, V82A, I50V, and V82F. By way of example, the residue referenced "VAL82" becomes the residue referenced "PHE82" when the mutation referenced "V82F" occurs. A target can have at least one, more than one or multiple mutations. HIV proteases exhibiting up to 40 mutations or even more (nearly one third of the residues) have been found in certain infected patients.

On the basis of experimental tests performed in vivo or in vitro, it is possible to draw up correspondence tables between non-mutated and mutated protease binding sites of HIV-1 and information relating to resistance against certain inhibitors (or drugs).

These tables are well known to the person skilled in the art, but they only provide information concerning a unique mutation and not several mutations. These tables are stored on paper media, but generally can also be stored in memories that can be accessible by the Internet site of the

University of Stanford so as to be used in computer software, for example the software known as "GRANT" (for Genotypic Resistance Analysis Tool) from the French company Infobiogen. More precisely, by supplying that software with the sequence of a mutation listed in one of its tables, its affinity parameter (resistance) relative to a known substance is output, when such a parameter exists. Those tables also contain information based on relatively poor statistics and which, furthermore, are not updated frequently and also do not contain standardized criteria. Also, the tables are generally used only once the phenotype of the patient's protease binding site of HIV-1 is known. These mutations are difficult to anticipate and phenotyping results are difficult to use. Moreover, phenotyping methods are based on local development, are non- standardized and fastidious, which renders them non-reproducible. Also, the number of operations implicated in phenotyping increase the number of unusable results on the order of 30%. Presently it takes several weeks for phenotyping, which can result in the patient's protease binding site of HIV-1 to further mutate. Since it is difficult to forecast the mutations, the results of the phenotyping become unusable.

In a different context, there is no method of calculating independently of environmental factors associated with the experimental tests in order to determine affinity data suitable for "quantifying" the capacity of a ligand (for example, a drug) for associating with a target, whose composition or structure has recently been identified, at a selected receptor site.

The observation whereby a pharmacological or therapeutic effect is the result of a specific interaction between a target including a receptor site and an effector (ligand), has led to the development of technologies enabling ligands to be identified experimentally. These technologies are effective on given targets. Particular mention can be given of the association of combinatorial chemistry which consists in generating an entire family of molecules from a base skeleton, and high throughput screening which consists in putting possible ligands, in particular those obtained by combinatorial chemistry, into the presence of one or more targets and measuring the magnitudes of the reactions that result from target/ligand association. Such technology is burdensome to implement, because of the quantities of chemical and/or biological materials that need to be used in parallel, and because of the complexity of the apparatus suitable for performing and interpreting the experiments. Thus, it is an object of the present invention is to provide a solution to the drawbacks described above, and more particularly to make it possible to estimate the capacity of a target for interacting with a ligand at least at one selected receptor site, even before a possible stage of experimental validation.

It is another object of the present invention to achieve considerable savings in time and in the number of target structures that need to be tested experimentally to identify a candidate ligand for development into a drug; for example to choose a more efficient drug for a mammal and especially when the mammal is infected with a virus.

It is yet another object of the present invention to provide a method of determining drug resistance.

It is yet another object of the present invention to provide a method for monitoring a mammal or human who is becoming resistant to drug treatment. It is yet another object of the present invention to provide a medical doctor with information very quickly such that the doctor can select a drug that is appropriate for a particular patient.

It is yet another object of the present invention to provide a method for determining the affinity value between a ligand and a target. These and other objects are achieved by the present invention as evidenced by the summary of the invention, description of the preferred embodiments and the claims.

SUMMARY OF THE PRESENT INVENTION

To this end, the invention provides apparatus for determining data concerning the interaction between a ligand and a target, wherein said apparatus comprises:

(i) a calculator means capable of receiving information representative of a target and of a ligand that interacts with the target at one of its receptor sites, and also capable of determining an affinity (or association) value representative of the association energy between said target and said ligand; and

(ii) a comparator means delivering data representative of an affinity level between the target and the ligand on the basis of the target/ligand affinity value determined by the calculator means, and on the basis of a selected reference parameter.

In another aspect the present invention provides a process for determining the affinity value between a target and a ligand.

In yet another embodiment, the present invention provides a method for determining drug resistance.

In another embodiment, the present invention provides a method to monitor mammals who are becoming resistant to drugs.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig.1 is a diagram illustrating the apparatus of the invention integrated in an Internet type communications network.

Fig. 2 is a block diagram illustrating the main constituent parts of apparatus of the invention. Fig.3 is a flow chart showing the main steps of a method of the invention.

Fig.4 is a continuation of the flow chart of Figure 3 showing the main steps of a method of the invention.

Fig. 5 shows the results of a retrospective study on data obtained in the Viradapt clinical trial. The ordinate corresponds to the calculated association energies. The bars show the intervals of obtained values in cases of success and failure, at 3 and 6 months.

Fig. 6 shows a validation test comparing Amprenavir resistance expected from GenMol calculations and effective resistance measured by Virco, for a great number of mutated HIV proteases. The x-axis corresponds to the calculated association energy, representative of the expected resistance, and the y-axis corresponds to the effective resistance. Fig. 7 shows a validation test comparing Amprenavir resistance expected from GenMol calculations and effective resistance measured by Virologic, for a great number of mutated HIV proteases. The x-axis corresponds to the calculated association energy, representative of the expected resistance, and the y-axis corresponds to the effective resistance(Regeneration logistic analysis).

Fig. 8 shows a validation test comparing Ritonavir resistance expected from GenMol calculations and effective resistance measured by Virco, for a great number of mutated HIV proteases. The x-axis corresponds to the calculated association energy, representative of the expected resistance, and the y-axis corresponds to the effective resistance(Regeneration logistic analysis).

Fig. 9 shows a validation test comparing Ritonavir resistance expected from GenMol calculations and effective resistance measured by Virologic, for a great number of mutated HIV proteases. The x-axis corresponds to the calculated association energy, representative of the expected resistance, and the y-axis corresponds to the effective resistance(Regeneration logistic analysis).

Fig. 10 illustrates the compensatory effect of the L10I mutation on the G48V mutation of the HIV protease, having respect to the resistance to

Ritonavir. The two columns represent the association energy between the protease and the Ritonavir calculated by GenMol, for a protease being mutated either once (G48V) or twice (L10I + G48V).

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The term "target" is used herein to mean any pathogen of a human, animal, or plant organism; it could equally well be (macro)molecules, pathogenic microorganisms such as viruses, bacteria, parasites, or fungi; or indeed cells infected by a virus or a parasite or prions. It can also be a tumor cell or a molecule that is specifically present at the surface of tumor cells. The term "ligand" is used herein to mean any type of structure capable of associating specifically with a receptor site of a target to block it or to reduce its activity, or indeed to modify its properties. It could thus be a chemical molecule or a chemical or biological macromolecule or an association thereof, or more generally any substance for therapeutic or prophylactic purposes such as medication for gene therapy or for cell therapy.

The term "site" is used herein to mean one or more locations on a target which defines a zone or a region with which a ligand can interact. Interaction comprises any type of association or bonding capable of modifying the physico-chemical or biological parameters or properties of the target.

The term "modeling" is used herein to mean determining by calculation certain physico-chemical or biological properties or parameters of targets or of entities formed by an association between a target and a ligand.

The term "mammal" as used herein means vertebrate animais, the females of which secrete milk for nourishing their young and include humans, dogs, cats, monkeys and the like.

Furthermore, the term "information representative of a target or of a ligand" means either the chemical or biochemical constitution of the target or the ligand, or else a structural and energy description thereof, or indeed merely their names (in which case a library of targets and ligands is used in which it is possible to find details about the constitution of the target or the ligand).

More specifically, the apparatus of the present invention comprises a calculator means and a comparator means. The calculator means are capable firstly of modeling (or generating or indeed building) not only the target and/or the ligand specified by the information received, but also an entity as constituted by associating a target with a ligand, and then of extracting a value from the model of the entity, which value is representative of the association energy between the target and the ligand.

The modeling is a result in particular of calculating free energy and bonding or non-bonding energy. It should thus be understood that the term "extract" is used herein to mean the operation which consists in seeking selected results from the numerous calculations performed during the modeling. Furthermore, the term "association" should be understood broadly. It covers any type of proximity interaction and does not necessarily require one or more physical bonds between the target and the ligand at a receptor site.

The apparatus of the invention is thus designed to determine an affinity value, preferably in the form of a target/ligand association energy, on the basis of biological or physico-chemical parameters of the target/ligand entity, and then to make a comparison with a reference in order to deduce affinity data which characterizes the aptitude of the ligand for interacting with the target.

The reference considered in order to deduce affinity data which characterizes the interaction between the ligand and the target is preferably a target/ligand entity for which experimental affinity data are known. The "reference parameter" is then said experimental affinity data. Depending on the context, it can be a resistance value, a dissociation constant, or any other measured parameter representative for the association between the target and the ligand. The reference parameter selected for performing the comparison can be the same for a series of target/ligand entities of a given type, or it can vary from one entity to another when experimental results are available, e.g., for reference entities of a type that is the same as or similar to the entity in question. For example, if the target is a mutated HIV protease and the ligand is the Ritonavir, the experimentally measured affinity of the Ritonavir for the wild-type HIV protease can be chosen as a reference parameter, whichever the mutations of the considered protease. Alternatively, if reliable experimental affinity data are known for the interaction between Ritonavir and a first mutated protease which is closer to the mutated protease under examination than the wild-type protease (for example, if said first mutated protease has 4 mutations and the examined protease has the same 4 mutations and 2 additional ones), the data concerning said first mutated protease can be chosen as reference parameter to deduce the affinity which characterizes the aptitude of Ritonavir for interacting with the examined mutated protease.

Furthermore, the reference parameter can be a numerical value or a range of numerical values. This can be preferable, particularly when the affinity parameter or reference parameter is the result of only a small number of experimental tests so that the statistics are not very reliable.

By means of the invention, it is thus possible to avoid numerous in vivo and in vitro tests and also to avoid the influence of the environment on the values of the affinity parameters, and consequently to obtain very quickly and at reduced cost interaction (or association) information that is reliable and not biased, are reproducible, are standardized and without delay.

Depending on requirements, the comparator means can deliver interaction (or association) data in the form of a warning message or in the form of the difference or the ratio between the selected reference parameter and an affinity parameter deduced, in particular, from the determined affinity value.

The term "warning" is used herein to mean a message of the type "poor association" or "strong resistance" which corresponds, for example to a comparison ratio giving a result greater than 1 (one); "very good association" or "poor resistance" which corresponds, for example, to a comparison ratio that is less than 1 (one); and "possible association" or "possible resistance" which corresponds, for example, to a comparison ratio of about 1 (one).

Consequently, affinity data is obtained that shows not only whether a ligand is capable of interacting (or associating) with a target, but above all this data permits the quantification of the quality of the interaction (or association). This information can be used by a furnisher who classifies ligand by order of increasing resistance, for example.

The apparatus may include other characteristics taken separately or in combination, and in particular:

- a memory for storing a plurality of multiplets (or n-tuplets) each comprising at least information representative of a target, information representative of a ligand, and a reference parameter for this target/ligand association, and in this case the comparator means are arranged in such a manner as to respond to receiving a target/ligand affinity value from the calculator means by extracting from the memory the reference parameter which corresponds to the target/ligand association in order to compare them. It is then particularly advantageous for the memory to be capable of storing, in the form of a first table, first targets referred to as "reference targets" (e.g., because they have not been the subject of modification they are then referred to as "wild type targets"), in correspondence with lists of second targets which are the results of modifications to which the first or reference targets have been subjected. Under such circumstances, it is possible to determine the affinity data between a modified target (a second target) and a ligand by using the reference parameter either for the association of the second target and the ligand or a close ligand when that is known, or by using a measured parameter for the association of the ligand and another modified target close to said second target, or else for the association of the ligand and the reference target (or first target). To do this, the calculator means are arranged in such a manner as to extract from the memory the first target which is associated with the second target they have received, e.g., together with a ligand, and then to determine first and second affinity values representative respectively of association energies between the first target and the ligand, and between the second target and the ligand, and then to take the difference between these first and second affinity values. Naturally, the comparator means then need to be arranged so as to deliver data representative of an affinity level between the second target and the ligand on the basis of the result of the difference taken by the calculator means and on the basis of the selected reference parameter, which parameter is preferably that which corresponds to the association between the first target and the ligand;

- selector means capable of delivering to the calculator means information representative of the targets and the ligands. These can be constituted by man/machine interface means coupled to a software module designed to find information in the memory representative of a target and/or a ligand, as supplied by a distant or local operator. For example, when the operator supplies only the name of a second target, the selector means search the memory for information concerning the first reference target and all of the known ligands with which it can be associated, so as to make it possible to analyze all of the associations between the second target and the various ligands found;

- a service terminal provided with a communications interface coupled to the calculator means or to the selector means (when they exist) and to a public communications network (of the Internet type) or a private communications network (of the Intranet type), and capable of receiving from an interrogation terminal data representative of the selected target and/or data representative of the selected receptor site and/or data representative of the selected ligand. Naturally, when the apparatus does not have selector means, it is preferable for the interrogation terminal(s) to address directly the information representative of a target and/or a ligand. When the target is a nucleic acid or a protein, it is advantageous for it to be transmitted in the form of a sequence (genotype) and/or of its function (phenotype). Under such circumstances, it is advantageous for the memory also to be capable of storing, in the form of a second table, targets in correspondence with their genotypes and/or their phenotypes. The targets present in the second table are preferably identical to at least the first targets contained in the first table. The selector means are then arranged in such a manner as to use the second table to determine the selected target when they receive information representative of its sequence (genotype) and/or of its expression and its function (phenotype). Furthermore, the service terminal (which can define an Internet (or "web") site) is preferably arranged in such a manner as to enable data representative of the level of target/ligand interaction (or association) to be transmitted to the interrogation terminal via the communications network. As a result, affinity data can be obtained very quickly from an interrogation terminal, which data will provide assistance in selecting a ligand (e.g. a drug) adapted to the target of interest; - the calculator means and the comparator means can also be arranged in such a manner as to store affinity values in the multiplets in memory, optionally together with affinity data as it is determined, in correspondence with the associated target and ligand. Under such circumstances, the calculator means and/or the selector means (when they exist) are advantageously capable, in the presence of a target and a ligand, firstly of determining from the multiplets whether an affinity value corresponds to them, and if one does, secondly of extracting said affinity value so as to transmit it directly to the comparator means. This procedure makes it possible to save time since it avoids the need to repeat calculations that have previously been performed;

- the calculator means and/or the selector means (when they exist) can also be arranged in such a manner as to operate in scan mode, i.e. following reception of information representative of a target, the second table is scanned through in order to determine all of the ligands which are associated therewith (or all of the ligands which are associated with the reference target of which the received target constitutes a modification), and then the affinity value is determined for the received target with each associated ligand; and

- a statistical calculator module enabling the value of an affinity parameter or a reference parameter stored in the memory to be modified, e.g. by statistical averaging, when it receives data from a user, preferably experimental data about the reference parameter and differing from the reference parameter as stored.

As mentionned above, the invention preferably makes use of 2 tables, namely :

• Table 1 , which establishes the correspondence between the reference targets (or first targets) and lists of second targets which are the results of modifications to which the first targets have been subjected. A typical element of this table is a wild-type protein put in correspondence with all the known mutated forms of this protein.

• Table 2 comprises a list of targets including at least all of the "reference targets" of Table 1 , put in correspondence with their genotypes and their phenotypes. The term "genotype" of a target means here any information representative of its sequence, whether it be its nucleotide or amino acid sequence. Throughout this application, the term "phenotype" should be understood broadly, as encompassing a wide range of kind of information, generally obtained through experiments and/or clinical observation, and concerning the target and, when available, its interactions with one or more ligands. For example, the resistance of a target to a ligand is part of the phenotype of said target, which will be advantageously considered in several applications of the present invention.

The invention also provides a method of determining affinity (or association) data between a ligand and a target, which method comprises the following steps: a) selecting a target and at least one ligand suitable for interacting with the target and one of its receptor sites; b) determining an affinity value representative of the association energy between the target and the ligand; and c) delivering data representative of an affinity (or association) level between the target and the ligand from the target/ligand affinity value and a selected reference parameter.

The method may include other characteristics taken separately or in combination, and in particular:

- in step a) it is possible to select the target from a plurality of multiplets each including information representative of a target, information representative of a ligand, and a reference parameter for the target/ligand association;

- in step c) (or indeed at the end of step b)), after a target/ligand affinity value has been received (or determined), the reference parameter corresponding to the target/ligand association is extracted from the memory in order to perform the comparison; - in a prior step, in particular prior to step a), it is possible to store a first table establishing correspondence between first targets referred to as "reference targets", and lists of second targets which are the result of modifications to which the first or reference targets are subject. It must be noted here that a second target can become a new reference target. For example, when sufficient experimental data are collected about a mutated target, this mutated target "MT" can become the reference for targets that are further mutated, but that share a greatest homology with the target "MT" than with the wild-type target. Under such circumstances, it is possible during step a) to select a second target and a ligand suitable for interacting with said second target, and then to extract from the first table the first target which is associated with said second target. Then, during step b) the first- and second affinity values are determined that are representative of the association energies respectively between the first target and the ligand, and between the second target and the ligand, and then the first affinity value is subtracted from the second affinity value. Finally, during step c) the data representative of the affinity level between the second target and the ligand is delivered on the basis of the result of the subtraction and on the basis of the reference parameter which is preferably the affinity parameter between the extracted first target and the ligand. Preferably, during step d), an affinity function is applied to the result of the subtraction in such a manner as to deliver an affinity parameter, and then the affinity parameter is compared with the selected reference parameter in order to determine the affinity data; this "affinity function" can be, for example, an empirical relationship between the calculated association energy and a measurable affinity parameter, based on measurements performed with a large number of targets;

- in step a) the target and/or the ligand can be transmitted by an interrogation terminal over a public or private communications network. Under such circumstances, it is advantageous for the target to be addressed in the form of a genotype and/or a phenotype, and for the target which corresponds to the genotype and/or the phenotype to be determined by interrogating a 1 o

second table that draws up correspondences between targets and their phenotypes and/or genotypes;

- in step a) the target and/or the ligand can be supplied by an interrogation terminal via a communications network selected from public networks and private networks. Under such circumstances, it is advantageous to provide a step d) in which the data representative of the target/ligand affinity (or association) level is transmitted to the interrogation terminal via the communications network;

- in step b) each determined affinity value can be stored in the multiplets in correspondence with the associated target and ligand so as to constitute a database. Under such circumstances, a preliminary stage is provided in step a) for the purpose of verifying in this database whether an association value exists that is stored in correspondence with the selected target and ligand in order to pass directly to the comparison of step c); and - in step b), on receiving a selected target, it is also possible to interrogate the first table in order to extract the first reference target associated therewith, if any, and then to interrogate the second table in order to extract all of the ligands corresponding to said first target and/or to the already known associated second targets, and finally to determine for each extracted ligand its affinity (or association) value with the target. The method then moves on to step c).

Numerous applications can be envisaged for the apparatuses and methods presented above, and in particular those in which the targets are pathogenic agents, molecules, or infected cells. More precisely, the invention is particularly advantageous when the pathogenic agents are viruses such as the human immuno-deficiency virus of type 1 (HIV-1 ) and the hepatitis B virus (HBV), when the molecules are proteins, nucleic acids, or compounds thereof, and when the receptor sites are mutated sites. In these particular circumstances, the ligands are preferably synthesized chemical molecules, biological macromolecules, or combinations thereof.

The present invention is not limited to only viruses, but also encompasses other microbial agents, parasites, fungus, bacteria, prions and any ligand that has a target. Also encompassed is any agent different from a microorganism or part thereof, and which is the object of treatment with molecules (for examples, tumor cells or chemical poisons).

In the description below, reference is made to an apparatus and a method for determining affinity data between targets and ligands. More precisely, reference is made, purely by way of example, to a target such as the protease binding site of human immuno-deficiency virus (HIV-1) and to ligands suitable for inhibiting the protease binding site of HIV-1 by association so as to block the late stage of virus maturation. The invention firstly provides an apparatus which preferably comprises a computer terminal such as a server or a personal computer.

In a variant, the apparatus of the invention need not comprise a computer terminal, but can be implanted in or can be designed to be implanted in a computer terminal. Under such circumstances, it can be implemented completely in the form of software modules (or programs) that are implanted in or that are designed to be implanted in a fixed or removable computer hard disk, or floppy disk, or more generally any type of data medium capable of co-operating with a computer, possibly via an internal or external reader. The apparatus can thus be implemented in the form of software which, once integrated in a calculator (or computer) enables said calculator to deliver data concerning the affinity between a target and at least one ligand, either in the form of a combination of electronic circuits ("hardware"), e.g., implemented on a dedicated electronics card, and computer instructions organized in software modules, or else in the form of a personal computer terminal or a service terminal connected to a public or private network, and fitted with the above-mentioned software or the above-mentioned hardware/software combination.

In the example shown in Figures 1 and 2, the computer terminal is a server S defining an Internet site specifically dedicated to calculating target/ligand affinity data. The server S is thus connected to a public network (Internet). However, in another embodiment, the server can be connected to a private network of the Intranet type.

The server S firstly comprises a calculator module 1 serving to model (or calculate) the structure and energy conformation of a target or of a ligand. It is well known that (macro)molecules, in particular, are characterized by their structural differences, and more particularly by their differences of shape, of electron charge distribution, and of nucleic charge distribution. More precisely, these electron and nucleic charge distributions contribute strongly to long-range electrostatic interactions which govern associations between a target and a ligand at a preferred site.

Numerous models have been developed for determining the free energy of a target, and more precisely the free energy minima that make it possible to quantify the degree of stability of a "system".

The calculator module 1 of the apparatus of the invention is arranged to perform such energy minimizations (e.g., in application of the laws of molecular dynamics), in such a manner as to determine the optimum configuration for a target such as the protease binding site of HIV-1 (or more generally a molecule or a macromolecule) on the basis of its molecular composition. The modeled target is thus the target possessing the most probable energy configuration, i.e., the minimum energy configuration, amongst all possible configurations.

Various software programs are known to the person skilled in the art for performing this type of modeling. Reference can be made in particular to the following: MIS, WOOLFORD, MODELLER (or LIGPLOT), and GENMOL. More precisely, the WOOLFORD program described in document

WO 98/54665 is designed to identify the location of ligand binding sites in proteins; design ligand molecules with optimal binding affinities for the selected target site; and refine lead compounds by defining the location and nature of chemical groups for optimal binding affinity. This algorithm requires knowledge of the high resolution structure of the protein but no knowledge of the location or identity of natural binding sites or ligands. Binding targets in the protein are identified and classified according to their expected optimal affinities. Binding targets can be located at the protein surface or at internal surfaces that become exposed as a result of partial unfolding, conformational changes, subunit dissociation, or other events. The entire protein is mapped according to the binding potential of its constituent atoms. The GENMOL program is designed so as to determine the most probable structure of a target (such as a (macro)molecule) on the basis solely of its constituents, even when not given in order (for example, when the target is constituted of several molecules), and the most probable configuration of an entity constituted by associating a target and a ligand. This modeling software can be obtained from the web site having the address http://chimir1.univ-mrs.fr or directly from the Organic Materials and Chemistry Laboratory of the Luminy Science Faculty at the following address: Laboratoire de Chimie et Materiaux Organiques, Moderation (LCMOM), Faculte des Sciences de Luminy, Case 901 , 163, Avenue de Luminy, 13288 Marseille Cedex 09, France and is publicly available from Mr. Gerard Pepe.

The calculator module 1 of the invention is also capable of extracting free energies from models of a target, of a ligand, and of a target/ligand entity in order to deduce an affinity value therefrom that is representative of the association energy of the target/ligand entity. As mentioned above, the term "association" is used herein to mean any type of direct or indirect interaction between a target and a ligand, at a main receptor site of the target (other or "secondary" interactions with auxiliary sites of the target can also be involved in the association). In other words, an association does not necessarily imply one or more bonds between the ligand and the residues of the target.

The affinity value calculated (or determined) by the calculator module 1 on the basis of knowledge of information representative of a target and a ligand is preferably the association energy between said target and said ligand. However, it could also be any other value representative of the association energy, i.e., related thereto by an equation or deducible therefrom by mathematical operations. As is known by the person skilled in the art, the equation of H. Eyring shows that the speed of association of a target with a ligand is related to the energy barrier that must be overcome to lead to target/ligand association.

This energy barrier is the target/ligand association energy E^ (in this case the

5 affinity value).

Furthermore, it results from Boltzmann statistics that the lifetime of a target/ligand entity, i.e., the length of time during which the target interacts with the ligand (or the duration of interaction, or indeed the duration of association), is related to the speed of target/ligand association. As a result 0 the lifetime of the target/ligand entity is directly related to the target/ligand association energy. At thermodynamic equilibrium, there is a relationship of the type given below: ^l09(^Dmutation association^/D reference association) (^E 1 ^" EAO)/ T = ΔE/RT ^wnere Re_ference association ^{is an affinitv} Parameter representing the duration of 5 association (usually referred to as "fold resistance") between a ligand and a reference target which is generally not mutated, D_{mutation association} is an affinity parameter representative of the duration of association between the ligand and a mutation of the reference target (or a mutated target), E_Ao represents the association energy of the reference target/ligand entity, and EAI 0 represents the association energy of the mutated target/ligand entity, T is temperature (in K), and R is the perfect gas constant (R = 8.314 J.mol^"1.K^'1 ≡ 2.10^"3 kcal.mol^"1.K^'1).

Preferably, the calculator means 1 determine the association energy between a target and a ligand on the basis of calculating the free energy of 5 said target and the free energy of the association of said target with the ligand. By convention, association energies are given in kcal/mole and have negative values, such that the free energy of a target/ligand entity is equal to the free energy of the target plus the free energy of the ligand, plus the energy for associating the target and the ligand (which is then a negative o value). This latter can then be expressed as :

EA ⁼ G target/ligand " G target " G ligand By way of example, the association energy between non-mutated HIV- 1 protease (in this case the reference target) and ritonavir, calculated by means of a calculator module using the GENMOL software, is equal to about -92.1 kcal/mole, while for the M46I, I84V, V82A, V82T, and V82F mutations, the following association energies with ritonavir are obtained respectively: -92 kcal/mole; -90.2 kcal/mole; -89.1 kcal/mole; -897 kcal/mole; and -87.8 kcal/mole, approximately. These values are merely examples. All association energies between ritonavir (or any other ligand) and other mutations of the protease binding site of HIV-1 can be calculated, including when the mutation applies to a plurality of residues. For example, for the following mutations: G84V + L10I; V82A + I54V + M46I + A71V; and I54V + V82A, the following association energies with ritonavir are obtained respectively: -93.2 kcal/mole; -89.4 kcal/mole; and -89.2 kcal/mole, approximately. By way of another example, the association energy between the protease binding site of HIV-1 that is non-mutated (here, the target reference) and nelfinavir, calculated with the aid of a calculating module 1 used in the software GENMOL is about in the environment of -81.4 kcal/mole, while the combinations of the mutations of the protease binding site of HIV-1 such as (D30N + L10I +L63LI + A71T + I93LI13V + L15V + I62V), (D30N + N88D), (D30N + M36I + L63P + N88DI13V + E35D) and (D30N + L10F + L63P + V77I + N88DK14Z + 115V + R41K) one obtains association energies with nelfinavir of about -81.8 kcal/mole, -81.3 kcal/mole, -79.5 kcal/mole and -78.5 kcal/mole, respectively. It is important to note that the calculated values for the association energies have a degree of error of about ± 0.3 kcal/mole.

From the relationship given above, the ratio between the association durations of a mutated target and of the reference target from which it derives can be deduced: D mu ,ta,tion associ .a ,tion ID re ,ference associ .a,tion oc10^ΔE/RT

This ratio shows that small differences between reference target/ligand and mutated target/ligand association energies can give rise to large differences between reference target/ligand and mutated target/ligand durations of association.

Thus, for an association energy difference ΔE = (E_AI - EAO) of about - 5 kcal/mole, the resulting ratio of the duration of association between a mutated entity constituted by a mutated target and a ligand, and a reference entity constituted by a reference target (non-mutated) and the same ligand, is of the order of 10 ^{"5 RT}. Consequently, with R ≤ 2.10^"3 kcal.mol^"1.K^"1 and T ≡ 300 K, the duration of association of the mutated target with the ligand is 10^ΔE/RT times the duration of association of the reference target with said ligand, which in this case means that it is smaller by a factor of about 10⁸.

Thus, if the value for the duration (or speed) of association is known for the reference target and the ligand, it is possible to determine the value for the duration of association between the mutated target and the ligand. This duration of association between the reference target and the ligand can be obtained experimentally. Naturally, certain durations of association can also be known for mutated target/ligand entities.

Duration of association (like speed of association) is thus an affinity parameter making it possible to quantify the aptitude of a target for associating with a ligand. However other affinity parameters exist that are equivalent to duration of association, in particular the resistance of a target relative to a ligand. Indeed this is a parameter which is the most usually known for reference targets. It is obtained during experimental tests performed in vivo or in vitro.

Resistance can be related to ΔE by an empirical relationship of the type log R = A. ΔE + B, wherein A and B are constants that depend upon the ligand. This relationship is obtained for a given ligand on the basis of association tests performed on a few mutations of a reference target, and, for each of these mutated target/ligand associations calculating the corresponding ΔE. In the following Table, are mentioned as an example, the (known) resistance and the (calculated) association energy of non-mutated (wild type) and mutated protease binding site of HIV-1 , with ritonavir:

In this case, the constants A and B have a relationship as mentioned above with a value of +0.4651 and +/- 0.15. In other terms, the resistance R of mutated binding site of the protease of HIV-1 with respect to ritonavir are linked to the association energy ΔE by the relationship log R = 0.4651. ΔE +/- 0.15.

In the following Table, are mentioned as an example, the (known) resistance and the (calculated) association energy of non-mutated (wild type) and mutated protease binding site of HIV-1 , with nelfinavir:

In the present case, the constants A and B mentioned above have the value of + 0.3468 and + 0.29. In other terms, the resistance R of mutated binding site of the protease of HIV-1 with respect to nelfinavir are linked to the association energy ΔE by the relationship log R = 0.3468. ΔE + 0.29. It is important to note that the values for resistance mentioned above were obtained from scientific publications.

Knowing the relationship between resistance and ΔE for certain mutations coming from the same reference target, it is possible to apply it to other mutations of the reference target. It is then possible to estimate the resistance of each mutation of a given reference target relative to a given ligand.

But the affinity parameter as determined, e.g., the resistance of a mutation relative to a ligand, can be information that is difficult to use. Consequently, in the present invention, once the affinity parameter between a target and a ligand has been determined, it is preferable to compare the affinity parameter with a reference parameter. For example, for the resistance of a mutation relative to a ligand molecule, the reference parameter for comparison purposes is preferably the known resistance relative to that ligand of the reference target from which the mutation is derived. Comparison can be performed by means of a mathematical operation, e.g., of the subtraction or of the ratio type (but more complex operations could be envisaged), having a result of a kind suitable for indicating whether the target/ligand affinity under analysis is of good quality or not.

For example, it can arbitrarily be decided that a resistance equal to 1 (one) means a reference target/ligand association that is durable, i.e., effective over time and all the values of resistance that are below 1 (one) have an association that is more durable. In which case, if the comparison consists in taking the ratio between the resistance of the mutation over the resistance of the reference target, then it can be considered that any comparison result that is greater than 1 (one) means bad association or high resistance, while a comparison result that is less than 1 (one) means association that is very good or resistance that is very weak. This comparison is performed by a comparator module 3 coupled to the calculator module 1 and implemented in the form of a software module and/or electronic components (hardware). The result of the comparison is communicated in the form of affinity data of message type or of numerical value type.

The message is preferably a warning of the type: i) "poor association" or "probable resistance". Such a message corresponds to a comparison having a result which, in the above example, is greater than 1 ; ii) "very good association" or "very weak resistance". Such a message corresponds to a comparison whose result, in the above example, is less than 1 ; or iii) "good association" or "weak resistance". Such a message corresponds to a comparison whose result, in the above example, is about 1. Other more precise types of messages are also encompassed in the present invention. Also the message can indicate that the resistance is weak, medium or high, for example in reference to a threshold.

These messages, or other equivalent messages, can be displayed on the monitor of the interrogating user or printed on a printer. The data in the form of a numerical value is the raw result of the comparison operation. It can then be compared by the user with values stored in a table in correspondence with a message of the type given above, possibly together with additional information or warnings.

In a variant, additional information can accompany the reference parameter associated with a target/ligand entity. For example, the additional information could relate to a limiting value of resistance beyond which the target/ligand resistance is considered to be insuperable. Under such circumstances, a result below the limiting value indicates that interaction between the mutated target and the ligand has good probability of being of good quality, whereas a result that is greater than said limiting value means that the interaction between the mutated target and the ligand has strong probability of being of poor quality, or even impossible. In another embodiment, the memory means of the apparatus includes a correspondence table between values and messages, such that once the result of the comparison operation is known, the comparison module extracts the message corresponding to the value of the result of the comparison (possibly together with associated additional information) from said table and displays it on the monitor of the interrogating user or printed on the printer.

On reading the comparison data, a medical practitioner knows immediately whether or not a given ligand can be used for inhibiting a target.

The reference parameter can also be a range of values. This can be preferable, particularly when the affinity parameter or reference parameter is the result of only a small number of experimental tests so that the statistics are not very reliable. This applies in particular to HIV-1 protease, to HIV-1 reverse transcriptase, and to a virus such as the hepatitis B virus (HBV), for which affinity and reference parameters are the result of a small number of tests performed in vivo on sick patients, or else in vitro.

In other applications, in particular such as high throughput screening, the affinity and reference parameters are the result of in vitro tests that are reproducible on a large scale and that consequently make it possible to obtain values that are reliable. When a plurality of target/ligand entities are initially selected by an operator, the apparatus delivers as much data representative of affinity levels as there are entities, said data being accompanied by information specifying the associated target and ligand.

The affinity and reference parameters provided by an operator or calculated by the calculator module 1 are stored in storage means, preferably as multiplets each comprising at least data representative of a target and data representative of a ligand. The storage means preferably comprise a memory 2 having the multiplets stored therein in the form of a file, or in a table in a library of the apparatus. As soon as an association energy difference (ΔE) is calculated by the calculator module 1 , it is preferably also stored in the corresponding multiplet. Other data could also be stored in the multiplets as soon as calculated or known. Particular mention can be made of the association energy of the target with the ligand, thus making it possible to save time when calculating the association energy difference (ΔE).

Naturally, when for a given mutated target/ligand entity a reference parameter is available that comes from experimental tests, then it is the experimental reference parameter which is used and stored in the multiplet.

It is clear that the invention is not limited to the methods of comparison described above by way of example. Any other type of comparison based on affinity values, affinity parameters, and a reference parameter can be envisaged in the context of the present invention.

The calculator means 1 and the comparator means 3 can be completely implemented in the form of software modules (or programs) stored on a hard disk of the computer terminal, internally or externally, or on a floppy disk, or more generally on any type of data medium capable of co-operating with a computer, possibly via an internal or external reader.

The storage means preferably store a first table which establishes correspondence between reference targets and their known mutations (referred to as mutated targets). The memory means preferably also store a second table establishing correspondence between targets (whether reference targets or mutated targets) and ligands with which they are suitable for interacting.

As a result, when the apparatus has information only about a mutated target to be analyzed, without reference to a ligand, it is possible in the correspondence tables and the multiplets to find the ligand(s) suitable for interacting with the received target. Three cases can arise. In the first case, there is at least one multiplet giving the received target and a corresponding ligand. In this case, the ligands are immediately extracted and transmitted to the calculator module 1. In the second case, there is no multiplet giving the received target, but the target is to be found in the first correspondence table. The first correspondence table is then used to find the reference target which is associated with the received target, i.e., the target of which it constitutes a mutation, and then the second table is searched as are the multiplets for ligands associated with the reference target. In the third case, there is no multiplet giving the received target and the target is not listed in the first correspondence table. The calculator module must then determine the energy and structure constitution of the received target by modeling, and then use the contents of the tables and the multiplet to determine by comparison the reference target whose constitution is the closest to that of the received target.

All of these determinations of reference targets and ligands can be performed by a selector module 4, preferably coupled to the communications interface 5 of the computer terminal and to the calculator module 1. As a result, when the apparatus receives data specifying a target, the selector module 4 determines firstly the information representative of said target, and secondly one or more ligands suitable for interacting with the received target and possibly also the reference target associated with the received target. Once in possession of this information, the selector module 4 can transmit it to the calculator module 1 so as to enable it to determine one or more affinity values. Furthermore, when the selector module 4 finds a multiplet which corresponds to the received target, it extracts directly therefrom data for transmitting either directly to the comparator module 3 when the data includes the affinity value and the associated affinity parameter, so as to enable it to perform its comparison, or else, when the data comprises only the affinity parameter, to the calculator module 1 so as to enable the calculator module to compute its model of the entity and then determine the affinity value.

The selector module 4 is preferably also arranged to propose to the user of the computer terminal that the target and/or ligand(s) be selected by the user from lists displayed on the monitor and preferably stored in the memory 3 of the apparatus. The list can be constituted directly by the correspondence tables. In another embodiment, such selection can be governed by an auxiliary module coupled to the storage means and to the calculator means 1 and/or to the selector module 4. This situation is particularly advantageous in terms of processing time, given that all of the data characterizing a target and a ligand offered for selection are already known to the apparatus of the invention, thus avoiding any need for it to determine them, e.g., by executing a complete model.

When the user decides not to operate in selection mode, or when that option is not made available by the apparatus, the user must supply the apparatus of the invention with data characterizing the target, optionally also with data characterizing one or more ligands. This data may comprise sufficient information to enable the calculator module 1 to determine one or more affinity values for each target/ligand association corresponding to the request of the user. When the data is not sufficient, the selector module 4 must determine the data representative of the target and of the ligand(s) for interacting with the target from the information stored in the tables and the multiplets of the storage means. Naturally, when the information is not available, the calculator module 1 proceeds to determine information by executing a complete model. Selection can be performed remotely when the apparatus of the present invention is installed in a service terminal S (or site) to which the user terminal is connected over a public network (of the Internet type) or a private network (of the Intranet type). Under such circumstances, the user must initially connect with the site (and more precisely with the service terminal), and then transmit to the site from the user's own terminal, the data representative of a target and possibly also of one or more ligands, in such a manner as to obtain in return one or more items of affinity data.

With the protease binding site of HIV-1 , it is advantageous for the user to transmit data representative of the target to the interrogation terminal or to the service terminal, where said data is in the form of a genotype (for example a series of mutations) and/or a phenotype (for example, a clinical observation).

To this end, the apparatus of the invention preferably includes a third table establishing correspondence between targets, preferably those which are already stored in the first and second tables and the multiplets, and their genotypes and/or phenotypes. This enables the apparatus, on receiving genotype and/or phenotype data, to interrogate the second correspondence table in order to obtain information representative of the target and to transmit that information to the calculator module 1.

The present invention makes it possible for a user, e.g., a doctor, to address data representative of a target characterizing a patient (the data preferably being in confidential form, e.g., encrypted), so as to obtain in return either from the interrogation terminal or from the service terminal, affinity data characterizing the suitability of the target in question to interact with one or more known ligands. A doctor can thus find out very quickly whether a known drug (ligand) is suitable for interacting effectively with the target of the patient. The doctor can then select the drug (or ligand) which appears to be the most appropriate to that particular patient.

However, it is also possible to connect to the site (Internet) in order to obtain affinity data for a series of targets belonging to a given "family", e.g., the mutations of a reference target. Thus, a user can address a given protease (or target) to the apparatus of the invention so that it determines the list of mutations of the protease and, for each mutation, the inhibitor (or ligand) with which it has the highest level of affinity. In other words, the apparatus determines affinity data for each mutation and for each ligand, and it then determines which ligand presents the greatest affinity. Thereafter, the apparatus draws up a comparative table presenting each mutation with the ligand with which it presents the strongest affinity.

The apparatus of the invention can also be used to determine affinity data of different types for a given target and a series of ligands, and then determine for each affinity data type the ligand which presents the strongest affinity with the target. The result can then be presented in the form of a table.

The same operation can be effected for each mutation of a reference target, such as the protease binding site of HIV-1 , so as to make it possible to compare in tables associated with each of the affinity data types ligands presenting the strongest affinity with the various mutations. The affinity data can be associated with other parameters relating to physico-chemical properties of the target/ligand entity, such as, for example, indications of the combination of a ligand with one or more other ligands. The invention may also allow to determine combinations of ligands (drugs) capable of acting with efficiency with respect to a given mutated target.

The apparatus of the invention preferably also includes a statistical calculator module 6 enabling the value of a reference or affinity parameter stored in a multiplet to be modified. Such modification can be automatic, i.e., without user intervention, or it can be under manual control, i.e., on an order from the user or from a site supervisor.

The modification can take place on receiving a new reference or affinity parameter value. Depending on the instructions sent by the user, it can be constituted either by purely and simply replacing the preceding reference or affinity parameter with the new reference or affinity parameter, or else by statistical averaging for the purpose of refining the preceding parameter. When the reference or affinity parameter previously stored in a multiplet is to be refined, it is essential for the statistical calculator module 6 to know the history of values for said parameter so as to be able to give an appropriate statistical weight to the parameter that has been communicated thereto.

Furthermore, it is particularly advantageous on each occasion that a new target is modeled and associated with a ligand, that a multiplet be created to store information representative of said target and of said ligand, the target/ligand affinity value as determined, and the reference parameter used for comparison purposes. This avoids the need to recalculate the affinity value used and to look for the pertinent reference parameter, and consequently makes it possible to proceed directly with comparison for obtaining the affinity data as soon as the affinity value and the reference parameter have been extracted.

The statistical calculator module 6 is preferably implemented in the form of a software module (or program), optionally integrated in the calculator module 1 , and implanted in the fixed or removable hard disk of a terminal, a floppy disk, or more generally on any type of data medium capable of cooperating with a terminal, optionally via an internal or external reader. The apparatus of the invention can also be used to validate affinity values. The example described below also relates to proteins. It is assumed that the affinity value is the association energy. For clarity reasons, the following notations will be used in this example : structures obtained by 5 experimentation (for example, by crystallography) will be noted by a capital letter followed by a number, while the same structures, obtained by calculation, will be designated by the same letter, but primed, and followed by the same number. A protein 1 will hence be noted P1 when its structure was obtained by crystallography, and P'1 when it was obtained by calculation. Still for clarity 0 reasons, an entity constituted by an association of a ligand L1 and a target P1 will be noted C1 rather than E1, in order to avoid confusion with the association energy, EA- This latter will be followed by the couple for which it was calculated : E_A(P1/L1 ) is thus the association energy calculated for P1 and L1 obtained from the data bank, and EΑ(P'1/L'1 ) the association energy of the calculated 5 P'1 and L'1. When one of the members of the couple for which the association energy is determined is calculated and the other obtained from the data bank, the energy will be noted with a prime as well (ex : E'_A(P'1/L1), when the target is calculated and the ligand obtained by crystallography). The same convention will be used for the associated entities : if one of the associated molecules is o calculated, the entity (C.) will be primed.

Firstly, information defining a protein (referred to as P1 ) and a first ligand (referred to as L1 ) e.g., non-mutated HIV-1 protease associated with ritonavir is extracted from a Protein Data Base (PDB), for example that of the Research Collaboratory for Structural Bioinformatics (RCSB), which is 5 accessible over the Internet at the address http//www.rcsb.org. Preferably, structural information that has been obtained by crystallography is extracted. Then, the association energy E_A(P1/L1 ) is determined by the calculator module 1 for the "native" entity C1 constituted by associating P1 as extracted from the PDB with L1 as extracted from the PDB. o Then, it is necessary to verify that the results of the energy calculations obtained from the structural information determined by the calculator module coincide with the results of energy calculations obtained from structural information coming from the PDB.

Thus, using the calculator module 1 , the protein P'1 and the ligand L'1 are modeled, and the association energy E'_A(P'1/L'1 ) is determined for the entity C'1 constituted by associating the model P'1 and the model ligand L'1.

Thereafter, E'_A(P'1/L'1 ) is compared with E_A(P1/L1 ) and E_A(P1/L1 ) is stored in the multiplet associated with P1 and L1 , providing EΑ(P'1/L'1 ) is substantially equal to E_A(P1/L1 ). When E'_A(P'1/L'1 ) is significantly different from E_A(P1/L1 ), then E'_A(P1/L1 ) is stored in the multiplet associated with P1 and L1. Indeed, a great difference in the association value often reflects a poor crystallography resolution, or the fact that protein in the crystal has adopted a particular conformation , different from its conformation in solution.

In these cases, the association energy obtained with the modeled structures is more reliable. The same procedure is applied to determine the association energies between the ligand L1 and the various mutations M1 i of the protease

P1.

Thus, the calculator module 1 is used to model a first mutation of the protein P1 (written M'11 ), and the association energy E'_A(M'11/L1 ) is determined for the mutated entity C'11 constituted by associating the first modeled mutation M'11 and the ligand L1 as extracted from the PDB.

Thereafter, information is extracted from the PDB defining the mutated entity C11 , constituted by said association between the first mutation M11 and the ligand L1. Thereafter, the association energy E_A (M11/L1 ) is determined for the mutated entity C11 by the calculator module 1. E'_A(M'11/L1 ) and E_A(M11/L1 ) are then compared, and E_A(M11/L1 ) is stored in the multiplet associated with M11 and L1 when E'_A(M'11/L1 ) is substantially equal to E_A(M11/L1 ). When E'_A(M'11/L1 ) is significantly different from E_A(M11/L1 ), then E'_A(M'11/L1 ) is stored in the multiplet associated with P1 and L1 , for the same reason as mentioned above. The same procedure is then applied for the following mutation M12 and the ligand L1 , and then for all mutations M1 i of P1 listed in the PDB. It is then possible to determine the association energies of P1 and its mutations M1 i with a second ligand L2. The same is then done for another protease P2, and so on.

The invention also relates to a method of implementing the apparatus described above. The method comprises at least the steps mentioned below.

Initially, a target and at least one ligand for interacting with the target at one of its receptor sites are selected. The term "selected" should be understood broadly. It amounts to determining information representative of a target and of at least one ligand suitable for enabling the calculator means to model each target/ligand entity. Thereafter, for each selected target/ligand association, an affinity value is determined that is representative of the association energy between the target and the ligand (this can be merely the association energy). This determination is performed by calculation when the affinity value is not previously known, or by extraction from the multiplets. Finally, for each target/ligand association, data is delivered representative of an affinity level between the target and the ligand, on the basis of the previously determined affinity value and of a selected reference parameter.

The reference parameter preferably comes from the multiplets. However, in a variant, it could be a fixed parameter valid for targets of a given type, e.g., a reference target and all of its mutations.

Reference is now made to Figure 3 to describe in greater detail a more complete method of the invention.

In a step 100, target and/or ligand data is selected. The selection can be implemented either by supplying data representative of a target on its own, or of a target and of one or more ligands. However, as mentioned above, it can also consist in selecting from lists a target optionally accompanied by one or more ligands.

In a step 110, a test is performed to determine the nature of the data supplied. More precisely, it is determined whether the data contains information about a target only, or about a target and one or more ligands. If the test 110 indicates that only a target has been specified, then, in a step 120, a new test is performed to determine whether said target is listed in the first correspondence table of reference targets and mutated targets.

If the received mutated target is listed in the first correspondence table, then the method passes onto a step 130 which is described below. In contrast, if the target is not listed in the first correspondence table, then the method proceeds to a step 122 in which the target is modeled so as to determine its configuration both in energy and in structure terms (modeling stage). Then, in a step 124, the first correspondence table and the multiplets are searched for the reference target which corresponds to the mutated target that has just been modeled. The method then moves onto step 130.

In step 130, information representative of the reference target is extracted.

Then, in a step 140, a new test is performed to determine whether the received target is known in the multiplets. If the target is not known in the multiplets, then in a step 142, the second target/ligand correspondence table is searched for the ligand(s) suitable for interacting with the reference target extracted in step 130. When the received mutated target corresponds to the determined (or extracted) reference target, it can reasonably be assumed that the ligands having the reputation of interacting with the reference target constitute good candidates for interacting with the received mutated target.

Once the ligands have been determined, an affinity value Vi is calculated in a step 144 for each ligand Li. Values V1 , V2, ..., VN are thus obtained when N ligands are specified in step 142. In a step 146, the associated reference parameter(s) Ri (R1 , R2, ...,

RN) is/are extracted from the multiplets associated with the reference target and each of the ligands. Thereafter, the method moves onto a step 170 which is described below.

When, in step 140, the received mutated target is found in at least one multiplet, then the method moves onto a step 150 in which a test is performed to determine whether an affinity value Vi is to be found in the multiplet(s) concerned in addition to the reference parameter Ri. If the multiplet specified in step 140 does indeed include an affinity value Vi and a reference value Ri, then the method moves onto a step 160 in which said affinity value Vi and said reference parameter Ri are extracted, after which it moves onto a step 170. In contrast, if the multiplet specified in step 140 does not have an affinity value, then the method moves onto a step 152 in which said affinity value Vi is calculated, and then the reference parameter Ri is extracted from the multiplet and the method moves onto step 170. Naturally, if a plurality of multiplets are specified in step 140, then the affinity value Vi is determined for each specified ligand Li with the received mutated target and the associated reference parameter Ri is extracted.

The description now returns to the test performed in step 110. If the result of this test indicates that a target and one or more ligands has been selected in step 100, then a test is performed in a step 112 to determine whether one or more multiplets exist corresponding to the selected target and the selected ligand(s). If this test indicates that multiplets do indeed exist corresponding to the selected target and ligand, then the method moves onto previously described step 150. However, if no multiplet exists corresponding to the selected target and ligand, or if there are not as many multiplets as there are selected ligands, then the method moves onto a step 114 of searching the first correspondence table for a reference target associated with the selected (and/or received) target. Then, the method moves onto step 152 described above so as to determine an affinity value Vi and a reference parameter Ri for each selected target/ligand pair. The description now returns to step 170 in which a counter is set to the value i = 1 (to indicate that the first comparison is being performed in a series of N, when N ligands have been selected) and the reference target/ligand association energy (E_Ao) is subtracted from the association energy to obtain the mutated target/ligand (E_Aι) , thus giving ΔE = E_Aι - E_Ao • Thereafter, the affinity parameter Pi is determined from the relationship associating it with ΔE.

Then, in a step 180, the reference parameter Ri is compared with the affinity parameter Pi to determine the affinity data Di. As mentioned above, various methods of comparison can be envisaged providing they make use of a selected reference parameter and the affinity value calculated by the calculator module and representative of the target/ligand association energy.

Then, in a step 190, the value of the counter is incremented by unity (i = i + 1 ).

Then, in a step 200, a test is performed to determine whether other comparisons need to be made. If so, the method returns to step 170. Otherwise, the affinity data Di is delivered in a step 210, e.g. to the monitor of the interrogation terminal of the user and/or to its printer, or else directly when the apparatus of the invention is integrated in the user terminal, or else via the network when the apparatus of the invention is integrated in a service terminal.

Naturally, the method described above with reference to Figure 3 is only an example. The method could include numerous variants and/or options for refining various steps or indeed for accelerating processing.

The invention is described above with reference to a particular application in the medical field, i.e., determining an inhibitor for the HIV-1 protease binding site. However, the invention applies more generally to all target/ligand entities in which the ligand is capable of associating with the target, and consequently for which target/ligand association energies can be calculated. This applies in particular to certain inhibitors of HIV-1 reverse transcriptase which are capable of preventing it from acting in the RNA and DNA association stage and in the duplication stage. By way of example, mention can be made of molecules of the "TIBO" family which include association complexes, and in particular nevirapine (or BI-RG-587 or viramune), efavirenz (or DMP-226 or L-743,427 or sustiva) and delavirdine (or BHAP der or U90152S or rescriptor) which stick to the reverse transcriptase to prevent it from acting.

However the invention relates to numerous other fields, and in particular all those which imply molecules, infected cells of a human, animal, or plant organism, and pathogenic agents, as well as microbes such as parasites, fungus, other viruses, bacteria, transmitted non-conventional agents such as prions and all ligands directed against targets, and the like. Also encompassed is a foreign agent of a microorganism that is the object of treatment with molecules.

For example, the molecules could be proteins, nucleic acids, and complexes comprising proteins and nucleic acids, as well as sugars, combinations of sugar-lipid proteins and other compositions having a structural base of sugar, lipids, amino acids and/or nucleic acids that are alone or in combination, and also PNA (Peptide Nucleic Acids, for which information can be obtained from the web site having the address http://www.bostonprobes.com).

Furthermore, in the description, the targets are mutations of a reference target. However that is merely an example and need not be the case, specifically when the fields in question are combinatorial chemistry and high throughput screening, fields in which the active principles of drugs are identified.

Besides the apparatus, the present invention also relates to a method for determining drug resistance in a mammal. This method comprises:

(a) obtaining a nucleic acid sequence of a target from a mammal;

(b) measuring an affinity value between said target and a ligand which is a drug;

(c) obtaining data of an affinity level between said target and said ligand on the basis of the target/ligand affinity value and a selected reference parameter; and

(d) determining the resistance of each mutation of a given target relative to a given ligand.

The nucleic acid sequence of the target from a mammal can be obtained by any manner known in the art. Generally, any sample containing the target nucleic acid sequence whether it is DNA, or RNA or cDNA or mRNA can be obtained from the mammal. The sample may be obtained, for example from serum, urine, cerebral spinal fluid, hair, tissue and the like provided that the target is present in the sample. The target nucleic acid is then isolated according to those methods known in the art as set forth in Sambrook et al, Molecular Cloning, A Laboratory Manual, 2^nd edition Cold Spring Harbor Press (1989).

The methods of obtaining the affinity value, the data on the basis of the affinity value of the target and a selected reference parameter, as well as determining the resistance were all discussed in detail above with respect to the apparatus and thus these same methods also apply for measuring drug resistance.

It should be recalled at this point that in determining the resistance of a drug a comparison value of greater than 1 means high resistance, while a comparison value of less than 1 means weak resistance!

In another embodiment, the present invention relates to a method of monitoring a mammal who is becoming resistant to a drug. This method comprises: (a) determining the genotype of the target in said mammal;

(b) determining the resistance of this target relative to a given ligand which is the drug to which said mammal is becoming resistant;

(c) identifying other ligands (drugs) that can be used for this target;

(d) comparing the resistance values obtained with the target and the ligands identified in step c); and

(e) changing the drug administered to the mammal when a ligand was found for which the resistance of the target is lower. Alternatively, a method of monitoring a mammal who is becoming resistant to a drug comprises:

(a) determining the resistance of each mutation of a given target relative to a given ligand which is a drug;

(b) determining a comparison value to the resistance; and (c) changing said drug when there is high resistance.

It should be clear that in thess monitoring methods the details of how o achieve each of the steps were discussed extensively above with respect to the apparatus of the present invention and are applied in the same manner in these monitoring methods.

Furthermore, in these monitoring methods, the health care provider such as a physician or a veterinarian, after receiving the comparison value, can decide whether the mammal or patient needs to change the particular drug or drugs which are being taken. For example, if there is very high resistance to a drug, in more instances than not, the health care provider, being aware of this fact, will change to another drug.

The health care provider merely has to forward the nucleic acid sequence of the target and the ligand(s) or drug(s) that the mammal or patient is currently taking and the comparison value can be obtained by the health care provider in less than an hour.

Hence, the present invention provides a rapid method for monitoring drug resistance. In yet another preferred embodiment, the present invention relates to a method for discovering a drug comprising:

(a) measuring a first resistance value (R1 ) and/or a first ΔE value of at least one mutation in a target relative to a ligand;

(b) altering said ligand to obtain a second resistance value (R2) and/or a second ΔE value (ΔE 2) wherein said second resistance value (R2) and/or said second ΔE (ΔE 2) value is lower than said first resistance value (R1 ) and/or said first ΔE value (ΔE1 ). The ligand can be altered by any means whether chemically, for example, by addition and/or deletions, substitutions of different sugar moieties and the like described above.

In yet another preferred embodiment, the present invention relates to a method for changing a drug used to treat a mammal comprising:

(a) using a first ligand which is a drug to mutate a target nucleic acid sequence; (b) changing said first ligand which is a drug to a second ligand which is also a drug wherein said second ligand has a lower ΔE and/or resistance than said first ligand. The above procedure is generally performed when there is a second ligand that may be more effective for treating the mammal than the first.

In yet another preferred embodiment the present invention provides a report comprising affinity data between a ligand and a target, said report generated from the apparatus as described above. This report can be in any form such as written, computerized form and the like.

While the invention has been described in terms of various preferred embodiments, the skilled artisan will appreciate that various modifications, substitutions, omissions and changes may be made without departing from the scope thereof. Accordingly, it is intended that the scope of the present invention be limited by the scope of the following claims, including equivalents thereof.

The following examples will further illustrate the performance and advantages of the present application, without limiting its scope.

EXAMPLES

Example 1 : Correlation between the association energy calculated by GENMOL and the resistance measured by Virologic, for mutated HIV proteases and Ritonavir.

Virologic is a biotechnology company which tests drug resistance and susceptibility in viruses such as HIV, HBV and HCV, through a technology called PhenoSense™. More information about this company and the PhenoSense™ technology can be found at www . virologic . com. The following table shows the correspondence between the association energy calculated by GENMOL, for the Ritonavir and 19 mutated HIV proteases, and the resistance of said proteases to Ritonavir, as measured by Virologic. The first column identifies the sample tested, the second one corresponds to the association energy calculated by GenMol, and the third one to the resistance measured in vitro by the PhenoSense™ technology.

The results show that there is a good correlation between the calculated association energy and the experimental data. Indeed, it can be noted that apart from a few exceptions, the lower the association energy (in relative value), the lower the resistance.

It appears that the median association energy for HIV proteases susceptible to Ritonavir is -92.53, while the median value for resistant proteases is -90.23.

Example 2 : Correlation between the association energy calculated by GENMOL and the resistance measured by Virologic, for mutated HIV proteases and Amprenavir.

The same comparison as in Example 1 has been performed for the Amprenavir. The results, shown in the following table, again demonstrate a good correlation between the association energy calculated by GENMOL and the resistance measured by the PhenoSense™ technology. Susceptible

Median : -75.85

Resistant

Median :

-73.31

The median association energy for the HIV proteases tested susceptible to Amprenavir in this assay is -75.85, and -73.31 for proteases resistant to Amprenavir.

Example 3 : Retrospective considerations on the Viradapt clinical study.

In order to test the reliability of GenMol to predict the clinical resistance of the HIV protease to drugs, and illustrate the methods of monitoring a mammal who is becoming resistant to a drug, according to the invention, a retrospective study was performed on the data collected in a clinical trial, which was itself performed in 1997-98 to test the effects of ritonavir on non- responsive patients suffering from AIDS.

This clinical trial, called VIRADAPT, involved 51 patients, who were in therapeutic failure. Samples were collected from these patients at the beginning of the study and at some points of the study. Their viral load was measured, and their phenotype resistance was determined by Virco (LabCorp, www.labcorp.com). A viral load decrease lower than 1 Iog10 in a patient was considered as failure.

The data collected in this trial were here used to calculate the association energy between the HIV protease of the patients involved in the trial and ritonavir, in order to retro-predict their resistance to the treatment. A Wilcoxon's test was performed for distribution comparison.

The results, presented in Figure 5 and summarized in the following table, demonstrate the clinical evidence of genotypic guided treatment. Indeed, it shows a very high correlation between the calculated association energy (ordinate in Figure 5) and the clinical result (success or failure), either at 3 months (p<0.01 with the Wilcoxon test) or at 6 months (p=0.03) following the beginning of the trial.

Example 4 : Validation set.

Figures 6 to 9 show further validation tests for a great number of mutated HIV proteases. The x-axis corresponds to the calculated association energy, representative of the expected resistance, and the y-axis corresponds to the effective resistance, as measured either by Virco (Fig. 6 and 8) or by Virologic (Fig. 7 and 9). The considered drugs are Amprenavir (Fig. 6 and 7) and Ritonavir (Fig. 8 and 9).

For each of these validations, the value of R² is indicated, and shows the relevance of the correlation.

Example 5 : Fine study of mutation patterns of the HIV protease

A resistance test is usually performed on patients infected by HIV in order to determine which drug would be efficient for them, especially for those who have already undergone a failure in their therapeutic strategy. This resistance test can be based either on the sequencing of (part of) the patient's virus (genotype), or on an in vitro resistance test using different concentrations of a drug (phenotype). Sequencing is most frequently used, since it is less expensive and more rapid.

The virologist, or the clinician knowing the genotype of the virus then interprets the detected mutations. This interpretation is usually performed by using an algorithm that associates at first a resistance level to each mutation or group of mutations. A great number of mutations in the regions coding for the protease or the reverse transcriptase are thus known to lower the activity of one anti-HIV drug. All these algorithms consider that a number of mutations lead to a reduction of the drug activity, and that an accumulation of such mutations systematically worsen the resistance to the antiviral drug.

However, it appears that the effect of such mutations is not always additive, and that mutations can have a different role whether alone or in combination. Rose et al have suggested that two mutations leading to a certain level of resistance when only one is present in the HIV protease, can lead to a weaker resistance when they are both present (Rose, R. E., Y. F. Gong, et al. (1996). "Human immunodeficiency virus type 1 viral background plays a major role in development of resistance to protease inhibitors." Proc Natl Acad Sci U S A 93(4): 1648-53). This indicates that the usually used algorithms are not sufficient to perform the analysis of complex mutation patterns.

Using the apparatus and methods of the present invention, it is now possible to model each of the possible combinations, and to compare them in order to evaluate and improve the above-mentioned interpretation algorithms. This can also lead to an increased knowledge about the active site and mechanism of action of the protease.

Fig. 10 show the association energy between the protease and the Ritonavir calculated by GenMol, for a protease being mutated either once (G48V) or twice (L10I + G48V). The obtained results illustrate the compensatory effect of the L10I mutation on the G48V mutation of the HIV protease, having respect to the resistance to Ritonavir, which is coherent with the observations by Rose et al, and hence illustrates the relevance of the method.

The present invention hence enables to determine mutation by mutation the interactions between a virus infecting a patient and a drug. This 5 can be used for the comprehension of new molecules, and even for the theoretical definition of novel antiviral drugs.

A step-by-step analysis of the mutation profile of a patient, - V3I, 113V, L33F, S37N, L63P, I64L, V77I, V82F, L90M-, was performed using GenMol according to the invention. 0 The following table comprises the results obtained with each of the above mutations taken alone (n° 1 to 7), which are very similar. A already described, the mutations V77I, L90M and V82F are those which lead to the strongest resistance.

The combinations n° 8 to 11 aim at determining the role of the V77I 5 mutation. The combination V77I-L90M was more resistant than V77I-V82F. The mutation V82F, in combination with the mutation V77I, exhibited a lower association energy to ritonavir than the mutation V82F alone; it appeared that this mutation V77I can, in certain circumstances, compensate the V82F mutation. In such cases, the accumulation of mutations would not lead to an o increase in resistance.

The role of the mutation L63P was examined in cases n°12 to 14. This mutation is very frequent and can nearly be considered as a natural polymorphism. The same results were found for L63P-L90M as L90M alone, and the same for V82F. Number 14 constituted a new result with respect to the 5 present knowledge about the mutations L63P and L33F, which are usually considered as minor mutations, but here lead to a certain level of resistance when combined.

The role of the mutation L33F was examined in cases n°15 and 16. This mutation could possibly compensate the mutation V82F. o The result obtained with the combination n° 17 confirmed earlier results.

Cases n° 18 to 24 constituted an extension of the preceding results and confirmed that the accumulation of mutations leads to an increased resistance, but also that the V77I mutation can possibly compensate the V82F mutation, which is the main mutation responsible for viral resistance to Ritonavir.

Calculations of association energies between the HIV protease and Ritonavir_> for different mutations or groups of mutations in the HIV protease.

Claims

What is Claimed is:

1. An apparatus for determining affinity data between a ligand and a target, wherein the apparatus comprises: -(i) a calculator means suitable for receiving information representative of a target and of a ligand that might interact with said target at a receptor site, and for determining an affinity value representative of the association energy between the target and the ligand on the basis of said received information; and - (ii) a comparator means arranged to deliver data representative of an affinity level between the target and the ligand on the basis of the target/ligand affinity value and a selected reference parameter.

2. The apparatus according to claim 1 , wherein the calculator means are arranged to determine the association energy between a target and a ligand by calculating the free energy of said target and the free energy of association of said target with the ligand.

3. The apparatus according to claim 1 or 2, further comprising a memory suitable for storing a plurality of multiplets having at least information representative of a target, information representative of a ligand, and a reference parameter for the target/ligand association, and wherein said comparator means are arranged, on receiving a target/ligand affinity value, to extract from the memory the reference parameter corresponding to the target/ligand association so as to perform the comparison.

4. The apparatus according to claim 3, wherein at least some of the multiplets contain an affinity value selected from a group comprising a target/ligand association value, an association energy difference, a target/ligand association speed and a target/ligand association duration.

5. The apparatus according to claim 3, wherein the memory is suitable for storing a first table containing first targets, referred to as "reference" targets, in correspondence with lists of second targets that are the result of modifications to the first or "reference" targets.

5

6. The apparatus according to claim 1 , wherein the calculator means are arranged to receive information representative of second targets referred to as "modified" targets that are the result of modifications to a first or a "reference" target, and representative of at least one ligand that might interact 0 with said second target, to extract from the memory the first target that is associated with said second target, and then to determine first and second affinity values respectively representative of the association energy between the first target and the ligand, and of the association energy between the second target and the ligand, and to deliver the result of taking the difference 5 between said first and second affinity values; and wherein the comparator means are arranged to deliver data representative of an affinity level between the second target and the ligand on the basis of the difference and of a selected reference parameter.

0 7. The apparatus according to claim 6, wherein the reference parameter is an association parameter for the first target and the ligand.

8. The apparatus according to claim 6, wherein the calculator means are arranged to apply an affinity function to the difference to deliver an affinity 5 parameter, and in that the comparator means are arranged to compare said affinity parameter with the selected reference parameter.

9. The apparatus according to claim 8, wherein the comparison is an operation selected from the group comprising taking the difference between o and taking the ratio of the first affinity parameter and the reference parameter.

10. The apparatus according to claim 6, wherein the data representative of affinity level is proportional to the result of said operation.

11. The apparatus according to claim 1 , wherein the data representative of the target/ligand affinity level is information selected from a group comprising the following warnings: "poor association" or "strong resistance", corresponding to a comparison ratio result greater than 1 (one); "very good association" or "weak resistance", corresponding to a comparison ratio result less than 1 (one); and "possible association" or "possible resistance", corresponding to a comparison ratio result of about 1 (one).

12. The apparatus according to claim 1 , further comprising selector means suitable for delivering information representative of targets and ligands to the calculator means.

13. The apparatus according to claim 1 , further comprising a service computer terminal fitted with a communications interface coupled to the selector means and/or to the calculator means, and also to a communications network selected from public networks and private networks, and suitable for receiving from an interrogation terminal data representative of the selected target and/or data representative of the selected ligand.

14. The apparatus according to claim 14, wherein said apparatus is suitable for being implanted in a service computer terminal fitted with a communications interface for supplying the selector means and/or the calculator means with information, and also with a communications network selected from public networks and private networks and suitable for receiving from an interrogation terminal data representative of the selected target and/or data representative of the selected ligand.

15. The apparatus according to claim 13, wherein the service terminal is arranged to transmit the data representative of the target/ligand affinity level to the interrogation terminal via said communications network.

16. The apparatus according to claim 15, wherein the service terminal defines a site that is accessible by the public Internet network.

17. The apparatus according to claims 1 , wherein said apparatus is suitable for being implanted in a personal computer terminal.

18. The apparatus according to claim 1 , wherein the data representative of the target is supplied in the form of a sequence and/or a phenotype.

19. The apparatus according to claim 3, wherein the memory is also arranged to store a second table containing the targets in correspondence with their sequences and/or their phenotypes, said targets being identical to those stored in the first table, and in that the selector means are arranged on receiving sequence and/or phenotype data to extract information from the second table that is representative of the target and suitable for the calculator means.

20. The apparatus according to claim 1 , wherein the calculator means are arranged to feed an affinity value accompanied by data representative of the corresponding target and ligand to the multiplets stored in the memory.

21. The apparatus according to claim 20, wherein each affinity value stored in a multiplet of the memory is representative of a target/ligand association energy.

22. The apparatus according to claim 12, wherein the selector means are arranged in the presence of a target and a ligand to determine whether there exists an affinity value stored in the multiplets of the memory corresponding to said target and to said ligand, and when such an affinity value does exist, to extract it and transmit it to the comparator means and/or the calculator means.

5 23. The apparatus according to claim 2, wherein the calculator means are arranged on receiving a second target to determine the first target and the ligands associated with said first target in the memory, and then to extract the reference parameter for each first target/ligand association.

0 24. The apparatus according to claim 1 , further comprising a statistical calculation module suitable for receiving a reference parameter from a user and corresponding to a target/ligand association, and for modifying the reference parameter in the multiplet corresponding to said target/ligand association as stored in the memory in the event of said received parameter 5 differing from the stored reference parameter.

25. A method of determining affinity data between a ligand and a target, the method comprising: a) selecting a target and at least one ligand that might interact with the o target at a receptor site; b) determining an affinity value representative of the association energy between the target and the ligand; and c) delivering data representative of an affinity level between the target and the ligand on the basis of the target/ligand affinity value and a selected 5 reference parameter.

26. The method according to claim 25, wherein in step a) the target is selected from a plurality of multiplets including at least information representative of a target, information representative of a ligand, and a 0 reference parameter for the target/ligand association; and in step c), on receiving a target/ligand affinity value, the reference parameter corresponding to the target/ligand association is extracted from the memory in order to determine said affinity data.

27. The method according to claim 26, wherein a first table is stored in the memory establishing a correspondence between first targets referred to as

"reference" targets, and lists of second targets that are the result of modifications to the first or "reference" targets.

28. The method according to claim 25, wherein in step a), a second target and a ligand reacts with said second target are selected initially, and then the first target associated with said second target is extracted from the memory; in that in step b) first and second affinity values are determined that are representative of the association energies respectively between the first target and the ligand and between the second target and said ligand, and then the difference is taken between said first and second affinity values; and in that in step c) data is delivered representative of an affinity level between the second target and the ligand on the basis of the difference and of a selected reference parameter.

29. The method according to claim 28, wherein the reference parameter is the first target/ligand association parameter.

30. The method according to claim 25, wherein in step c), the difference is applied to an affinity function so as to deliver an affinity parameter, and then said affinity parameter is compared with the selected reference parameter in order to determine said affinity data.

31. The method according to claim 25, wherein in step a) the target and/or the ligand are transmitted by an interrogation terminal over a communications network selected from public networks and private networks.

32. The method according to claim 31 , wherein the target is addressed in the form of a sequence or a phenotype, and in that in step a), the target corresponding to the received sequence or phenotype is determined from a second table stored in the memory and establishing correspondence between targets and their corresponding sequences and/or phenotypes.

33. The method according to claim 25, further comprising a step d) in which data representative of the affinity level between the selected target and ligand is transmitted to the interrogation terminal over a communications network.

34. The method according to claim 25 or claim 26, wherein in step b) its determined affinity value is stored in the multiplet in correspondence with the associated target and ligand so as to build up a database.

35. The method according to claim 25 or claim 26, wherein in step a) the multiplets are interrogated to determine whether an affinity value is stored in correspondence with the selected target and ligand so as to pass directly to step c) when said affinity value already exists.

36. The method according to claim 25, wherein in step b), on receiving a selected target, the multiplets are searched for all associated ligands, and an affinity value is determined for each target/ligand association found, and in step c) the reference parameter of each found target/ligand association is extracted to proceed with comparisons of each affinity value with the corresponding reference parameter.

37. A method for determining drug resistance, said method comprising:

(a) obtaining a nucleic acid sequence of a target from a mammal; (b) measuring an affinity value between said target and a ligand which is a drug; (c) obtaining data of an affinity level between said target and said ligand on the basis of the target/ligand affinity value and a selected reference parameter; and

(d)determining the resistance of each mutation of a given target 5 relative to a given ligand.

38. The method of claim 37, wherein a comparison value is used in determining said resistance.

0 39. The method of claim 38, wherein a comparison value of greater than 1 means high resistance, while a comparison value of less than 1 means weak resistance.

40. The method according to claim 37, wherein said nucleic acid sequence of 5 said target is DNA or RNA or PNA.

41. A method of monitoring a mammal who is becoming resistant to a drug, said method comprising:

o (a) determining the resistance of each mutation of a given target relative to a given ligand which is a drug;

(b) determining a comparison value to the resistance; and

(c) changing said drug when there is high resistance.

5 42. The method according to claim 41 , wherein said resistance is determined based on an affinity level between said target and said ligand and a selected reference parameter.

43. The method according to claim 41 , wherein a comparison value is used in 0 determining said resistance.

44. The method according to claim 43, wherein a comparison value of greater than 1 means high resistance, while a comparison value of less than 1 means weak resistance.

45. The method according to claim 42, wherein said affinity level is determined based on measuring an affinity value between said target and a ligand which is a drug.

46. The method according to claim 45, wherein said target is a nucleic acid sequence.

47. The method according to claim 46, wherein said nucleic acid sequence is RNA or DNA.

48. The method according to claim 40 or claim 46, wherein said RNA is mRNA and said DNA is cDNA.

49. A method for discovering a drug comprising:

(c) measuring a first resistance value (R1 ) and/or a first ΔE value of at least one mutation in a target relative to a ligand;

(d) altering said ligand to obtain a second resistance value (R2) and/or a second ΔE value (ΔE 2) wherein said second resistance value (R2) and/or said second ΔE (ΔE 2)value is lower than said first resistance value (R1 ) and/or said first ΔE value (ΔE1 ).

50. A method for changing a drug used to treat a mammal comprising:

(c) using a first ligand which is a drug to mutate a target nucleic acid sequence; (d) changing said first ligand which is a drug to a second ligand which is also a drug wherein said second ligand has a lower ΔE and/or resistance than said first ligand.

51. A report comprising affinity data between a ligand and a target said report generated from the apparatus of Claim 1.