CA2259314A1 - Method and apparatus for alignment of signals for use in dna base-calling - Google Patents

Method and apparatus for alignment of signals for use in dna base-calling Download PDF

Info

Publication number
CA2259314A1
CA2259314A1 CA002259314A CA2259314A CA2259314A1 CA 2259314 A1 CA2259314 A1 CA 2259314A1 CA 002259314 A CA002259314 A CA 002259314A CA 2259314 A CA2259314 A CA 2259314A CA 2259314 A1 CA2259314 A1 CA 2259314A1
Authority
CA
Canada
Prior art keywords
data
traces
peaks
data traces
coefficients
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
CA002259314A
Other languages
French (fr)
Inventor
Rodney D. Gilchrist
Vrijmoed Chi
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Bayer Healthcare LLC
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Publication of CA2259314A1 publication Critical patent/CA2259314A1/en
Abandoned legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N27/00Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
    • G01N27/26Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating electrochemical variables; by using electrolysis or electrophoresis
    • G01N27/416Systems
    • G01N27/447Systems using electrophoresis
    • G01N27/44704Details; Accessories
    • G01N27/44717Arrangements for investigating the separated zones, e.g. localising zones
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16BBIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G16B30/10Sequence alignment; Homology search
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N27/00Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
    • G01N27/26Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating electrochemical variables; by using electrolysis or electrophoresis
    • G01N27/416Systems
    • G01N27/447Systems using electrophoresis
    • G01N27/44704Details; Accessories
    • G01N27/44717Arrangements for investigating the separated zones, e.g. localising zones
    • G01N27/44721Arrangements for investigating the separated zones, e.g. localising zones by optical means
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12Classification; Matching
    • G06F2218/14Classification; Matching by matching peak patterns
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16BBIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16BBIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00ICT specially adapted for sequence analysis involving nucleotides or amino acids

Abstract

Data traces from four channels of an automated electrophoresis detection apparatus are aligned by identifying peaks in each of the four data traces; optionally normalizing the data traces to achieve a uniform peak height; combining the four data traces in an initial alignment; and determining coefficients of shift and stretch for selected data points within each data trace. The coefficients are determined by optimizing a cost function which reflects the extent of overlap of peaks in the combined normalized data traces to which the coefficients have been applied. The cost function is optimized when the extent of overlap is at a minimum. The coefficients are then used to generate a warp function for each data trace. These warp functions are applied to their respective data traces to produce four warped data traces which are aligned to form an aligned data set. The aligned data set may be displayed on a video screen of a sequencing apparatus, or may be used as the data set for a base-calling process.

Description

METHOD AND APPARATUS FOR ALIGNMENT OF SIGNALS
FOR USE IN DNA BASE-CALLING
DESCRIPTION
BACKGROUND TO THE INVENTION
This invention relates to a method of processing output signals from an automated electrophoresis detection apparatus, and to an apparatus which employs this method for sequencing nucleic acids.
One of the steps in nucleotide sequence determination of a subject nucleic acid molecule is interpretation of the pattern of nucleic acid fragments which results from electrophoretic separation of fragments, or reaction products, of a DNA sequencing reaction (the "fragment pattern"). The interpretation, colloquially known as "base calling", involves determination from the recorded fragment pattern of the order of four nucleotide bases, A
(adenine), C (cytosine), G (guanine) and T (thymine) for DNA or U (uracil) for RNA in the subject nucleic acid molecule.
The chemistry employed for a DNA sequencing reaction using the dideoxy (or chain-determination) sequencing technique is well known, and was first reported by Sanger et al.
(Proc. Natl. Acad. Sci. USA 74: 5463-5467 (1977)). Four samples of nucleic acid fragments (terminating in A, C, G, or T(U) respectively in the Sanger et al. method) are loaded at a loading site at one end of an electrophoresis gel. An electric field is applied across the gel, causing the fragments to migrate from the loading site towards the opposite end of the gel.
During this electrophoresis, the gel acts as a separation matrix. The fragments, which in each sample are of an extended series of discrete sizes, separate into bands of discrete species in a lane along the length of the gel. Shorter fragments generally move more quickly than larger fragments.
If the DNA fragments are labeled with a fluorescent label, an automated electrophoresis detection apparatus (also called a "DNA sequencer") can be used to detect the passage of migrating bands in real time. Existing automated DNA sequencers are available from Applied Biosystems, Inc. (Foster City, CA), Pharmacia Biotech, Inc. (Piscataway, NJ), Li-Cor, Inc.
(Lincoln, NE), Molecular Dynamics, Inc. (Sunnyvale, CA) and Visible Genetics Inc.
(Toronto). Other methods of detection, based on detection of features inherent to the subject molecule, such as detection of light polarization as disclosed in US Patent No.5,543,018 which is incorporated herein by reference, are also possible.
A significant problem in determining a DNA sequence, encountered particularly with high speed DNA sequencing and in sequencing apparatus which do not combine the four sets of sequencing reaction products in a single lane, is alignment of data signals from the four different output channels of an automated DNA sequencing apparatus. Once data is aligned properly, it is relatively straight-forward to base-call it. However, this initial step can be very challenging since the output signal may be erratically shifted and/or stretched as a result of chemistry and gel anomalies. A reliable method of aligning data, that can produce data which takes into account non-linear shifting and stretching of signal output, is highly desirable particularly for high-speed DNA sequencing.
Existing prior art determinants in this field are very limited. Existing automated sequencers traditionally operate at voltages low enough that non-linear shifting is avoided.
The use of low voltages, however, limits the speed with which separation of sequencing fragments into discrete bands can be accomplished.
Published methods of computer assisted base calling include the methods disclosed by Tibbetts and Bowling (US Pat.No.5,365,455) and Dam et al(US Pat.No.5,119,316) which patents are incorporated herein by reference. Both patents assume alignment of output signals and address only aspects of base-calling from the aligned signals.
It is an object of the present invention to provide a method of aligning real-time signals from the output channels of an automated electrophoresis apparatus.
It is a further object of the invention to provide an improved method of base-calling an DNA signal sequence aligned according to the invention.
It is still a further object of the invention to provide an apparatus sequencingnucleic acids which utilizes the improved method in accordance with the invention for aligning real-time signals from the output channels of an automated electrophoresis apparatus.
SUMMARY OF THE INVENTION
These and other objects of the invention are achieved using a method for aligning data traces from four channels of an automated electrophoresis detection apparatus, each channel detecting the products of one of four chain-termination DNA sequencing reactions, whereby CA 022~9314 1998-12-22 WO 98/00708 PCT/CAg7/00463 said four channels together provide information concerning the sequence of all four bases within a nucleic acid polymer being analyzed, comprising the steps of:
(a) identifying peaks in each of the four data traces;
(b) norm~li7in~ the height of said peaks in each of said data traces to a common value to generate four norm~li7ed nata traces if the peaks are not of substantially equal height;
(c) combining the four normalized data traces in an initial alignment;
(d) deterrnining coefficients of shift and stretch for selected data points within each normalized data trace, said coefficients optimizing a cost function which reflects the extent of overlap of peaks in the combined normalized data traces to which the coefficients have been applied~ said cost function being optimized when the extent of overlap is at a minimum;
(e) generating warp functions for the normalized data traces from the coefficients of shift and stretch determined for the selected data points;
(f) applying the warp functions to the respective data trace or normalized data trace to produce four warped data traces; and (g) assemblin~ the four warped data traces to form an aligned data set.
The aligned data set may be displayed on a video screen of a sequencing apparatus. or may be used as the data set for a base-calling process.

BRIEF DESCRIPTION OF THE DRAWINGS
Fig. I shows the alignment process of the invention in flow chart form;
Fig. 2 illustrates a preliminary (unaligned) signal recorded from an automated DNA
sequencing apparatus over 15 seconds;
Fig. 3 illustrates the detection of peaks from a preliminary signal;
Fig. 4 illustrates a normalized data stream;
Figs. 5A and 5B illustrate a norrn~li7~tion method employed for peaks and valleys;
Figs. 6A and 6B illustrate the preliminary alignment of four norrn~li7ed data streams representing time based aliL~nment of each of the A, C, G and T output channels;Fig. 7 illustrates alignment of the 4 norm~li7ed data streams with respect to minimi7:~tion of the "cost" function;

SUBSTITUTE SHEET (R(UlE 26) CA 022~9314 1998-12-22 Figs. 8A and 8B shows exemplary warp functions for data points in the C, G and Ttraces relative to the A trace;
Fig. 9 illustrates a method for determining a standard gaussian peak width;
Fig. 10 shows an example of the classification of singletons using the features of peaks 5 in the detected signal;
Fig. 11 illustrates a base-calling method useful with the aligned data traces; and Fig. 12 shows an apparatus in accordance with the invention.

DETAILED DESCRIPTION OF THE INVENTION
The present invention provides a method of aligning data traces which can be used to align data traces from an automated electrophoresis detection apparatus for use in base-calling. In accordance with the invention. an optimization routine such as a simulated annealing algorithm is used to determine coefficients of stretch and shift to align data traces.
each representing one of the four chain-termination DNA sequencing reactions. With high quality data trace alignment, base-calling may proceed with a high degree of accuracy and repeatability.
"Data trace" as used in the specification and claims of this application refers to the series of peaks and valleys representing the migrating bands of oligonucleotide fragments produced in one chain termination sequencing reaction and detected in a DNA sequencer.
Such data traces are sometimes referred to as sequence chromatograms or a chromalographic trace. The data trace may be either a raw data trace or a "conditioned" data trace.
"Shift" as used in the specification and claims of this application is interchangeable with "offset" and refers to the number of data points which the signal output is displaced, either positively or negatively from its original position.
"Stretch" as used in the specification and claims of this application refers to increase or decrease in spacing between data points of a signal output relative to the original spacing. The stretch may be constant across a segrnent of data or may follow a second or higher order polynomial.
A "warp function" as used in the specification and claims of this application is an instantaneous representation of optimized shift and stretch at each data point. The warp function can be represented graphically either as a data point-for-data point plot for one data SUBSTITUTE SHEEr ~121JlE 26~

CA 022~9314 1998-12-22 trace versus a standard or second data trace~ or as a plot of displacement as a function of data point number. Standard traces may be a trace representing an average of multiple experimental runs, a base separation function, or a standard derived from a text sequence. The warp function is applied to a data trace to obtain a "warped data trace."
Fig. I shows the alignment process of the invention in flow chart form. As shown, the first step in the method is the identification of peaks in each of several (generally four) data traces. In some cases~ these traces are next norm~li7e~ to a constant height. This step may be unnecessary if the DNA Sequencer and chemistry used produce substantially uniform peak sizes (homozygous peak heights within about 10 %) or if the traces are nearly aligned from ~he sequencer. The terrn "normalized data traces" refers to data traces having this level of consistency, whether as a result of norm~ tion step or as a result of sufficient initial uniformity. Normalization may also be unnecessary~ even for non-uniform peak sizes~ if the cost function used in the alignrnent is independent of peak height and area.
The data traces are then combined in an initial alignment and divided into a plurality of segments or windows. For each segment, optimized stretch and shift values are determined using an iterative optimization routine which converges towards a "best" solution for alignment of the data traces. The quality of the various trial alignment is assessed for comparison purposes using a cost function. Various cost functions can be employed including ~1 ) a cost function which minimi7es overlap of the data traces combined within the segment;
and (2) a cost function which maximizes the regularity of peak distribution the segment.
Once a "best'' solution has bene found, the shift and stretch values are used to define warp functions. These warp functions are applied to the individual data traces to produce warped data traces which are assembled as aligned data useful for base-calling.
To illustrate the use of this method, Fig. 2 shows a typical raw data trace for one of the four chain-termination DNA sequencing reactions~ as detected on a Visible Genetics Inc.
MicroGene BlasterTM. The X-axis represents time while the Y axis represents fluorescence detection. The data trace reveals a series of bands of fluorescent molecules passing through the detection site, as expected from a typical chain-termination DNA sequencing reaction.
Fig. 2 shows several of the features which complicate the use of raw data for base-calling, and illustrates the need for the present invention to provide app,opliately aligned signals for this purpose. In particular, as reflected in Fig. 2, the fluorescence intensity can vary SUBSTITUTE SI~EET (RVLE 26) CA 022~9314 1998-12-22 from one band to another. In addition, not all bands are fi~lly resolved, and the spacing between adjacent bands is not always an integral multiple of the theoretical spacing between adjacent bands. The present invention provides a method for converting this raw signal, and its three counterpart signals for the other three sequencing reactions, into aligned data which is S highly suitable for base-calling.
The data trace which is processed in accordance with the method of the invention is preferably a signal collected using the fluorescence detection apparatus of an automated DNA
sequencer. ~owever, the present invention is applicable to any data set which reflects the separation of oligonucleotide fragments in space or time, including real-time fragment patterns using any type of detector, for example a polarization detector as described in US Patent No.
S,543,018; densitometer traces of autoradiographs or stained gels: traces from laser-scanned gels containing fluorescently-tagged oligonucleotides: and fragment patterns from samples separated by mass spectrometry.
In the method of the invention, four data traces, one for each sequencing reaction, are normalized as needed to correct for variations in peak height using the procedure described below. The normalized data traces are then used to deterrnine a series of stretch and shift coefficients, which are then applied to the norrnalized data traces to arrive at aligned data traces.
Prior to norm~ ing and aligning the data traces for the four sequencing reactions using the method of the invention, however~ it may be advantageous to condition the signal, although this step is not required. This conditioning can be done, for example~ using conventional baseline correction and noise reduction techniques to yield a "conditioned" data trace. As is known in the art, three methods of signal processing commonly used are background subtraction, low frequency filtration and high frequency filtration, and any of these may be used, singly or in combination to produce a conditioned signal to be used as a conditioned data trace in the method of the invention.
Preferably, the data is conditioned by background subtraction using a non-linear filter such as an erosion filter, with or without a low-pass filter to elimin~te systemic noise. The preferred low-pass filtration technique is non-causal gaussian convolution.
After any needed conditioning of the data trace is perforrned~ the data trace isnorrn~ ed as needed to generate a "norrnalized data trace" which is used to determine SUBSTITUTE SHEET (RULE 26) CA 022~9314 1998-12-22 wo 98/00708 PC rlCAg7l00463 coefficients of stretch and shift for base-calling. The normalization process includes the following steps.
Firstly, the data trace (raw or conditioned) is searched for peaks. Peaks can beidentified as the middle data point of three consecutive data points wherein the inside data point is higher than the two outside data points. Fig. 3. More sophisticated methods of peak detection are also possible. For example, a preferred method involves using the "three-point"
method to segrnent the data trace, and then joining the segments. A trace feature is assigned as an actual peak whenever the difference between a maximum and an adjacent minimum exceeds a threshold value, e.g., 5%. A minimum peak height from the base-line may also be required to e1imin~te spurious peaks.
An exception is made for the so-called "primer peak" and "terrnination peak" which are found in some variations of the chain-termination sequencing method. These peaks comprise a large volume of unreacted prim~r, which tends to interfere with base-calling around the shorter chain-extension products, and a large volume of the complete sequence which may interfere with base-calling around the longest chain-extension products. These peaks are identified and elimin~tecl from consideration either on the basis of their size, their location relative to the start and end of the electrophoresis process, or some other method.
After elimin~tion of the primer and termination peaks, the data trace is norrn~li7e~ so that all of the identified peaks have the same height which is assigned a common value, e.g., 1.
(Fig. 4). This process reduces signal variations due to chemistry and enzyme function. and works effectively for homozygous samples and for many heterozygotes having moderate, i.e Iess than about S to 10%, heterozygosity in a 200 base pair or larger region being sequenced.
It will be appreciated that when the data trace has substantially uniform peak height, norm~li7~ion can be omitted.
To norrnalize the data trace, the points between each peak are assigned a numerical height value based on their position in the data trace relative to a hypothetical line joining consecutive peaks and the base line of the signal. Thus, as shown in Fig. SA, the valley between peaks 1 and 2 has a minimum at a point which is approximately 25% of the distance from the baseline B to the line C joining peaks I and 2. The minimum of this valley is therefore assigned a value of about .25. (See Fig. SB). Similarly, the valley between peaks 2 and 3 has a minimum at a point which is approximately 80% of the distance from the baseline SUBStlTUTE SHEEr (RULE 26) ,~

CA 022~9314 1998-12-22 wo 98/00708 PCT/CA97/00463 B to the line D joining peaks 2 and 3. The minimum of this valley is therefore assigned a value of about 0.8 in the normalized data traee.
The next step in the method of the invention is the determination of shift and streteh coeffieients for a set of four normalized data traces, one for each se~uencing reaetion. This is S aeeomplished by combining the four norm~li7e~ data traees as shown diagrammatieally in Figs. 6A and 6B, and determining eoeffieients of shift and streteh for seleeted data points within eaeh norm~li7ed data traee which optimize a "cost" funetion. The cost funetion generally refleets the suitability of a trial alignment resulting from the applieation of the trial streteh and shift eoeffieients to the eombined data traees. One type of eost funetion evaluates 10 the extent of overlap of peaks in the eombined normalized data traees to whieh the eoeffieients have been applied, and is eonsidered optimized when the extent of overlap is at a minimum. It will be understood that the terms "best solution," "optimized" and "minimum" as used herein do not require absolute optimization to an absolute minimum, whieh eould requireunreasonably long periods of analysis time, but only require a praetieal level of optimization 15 sufficient to achieve satisfactory alignment of the data traces for base-calling.
In one embodiment, the cost function measures the total area above the combined normalized data traces, i.e., the dotted area in Fig. 5B. The norn~li7ed data traces are then shifted and stretched in an effort to minimi7.e the value of "cost." In a second embodiment, "cost" is set equal to the area below the combined normalized data traces, and the data traces 20 are then shifted and stretched to maximize the value of cost (i.e. reducing the overlap of the peaks). ~owever, it has been found empirically that this latter approach emphasizes less valuable features of the data traces than using the area above the curves as the eost funetion.
In another embodiment, the cost function combines the total area above the combined data traces (the dotted regions in Fig. 6B) with the area below the highest edge formed by the 25 combined data traces and above the second highest edge of the combined data traces (the lined regions in Fig. 6B). The cost function is at a minimum when the first of these areas is minimi7ecl and the seeond of these areas is maximi7e~ When the eoeffieients which produee the minimum eost are applied to the norm~li7ed data traces, an aligned nonn~li7e~i data set as shown in Fig. 7 results.

SUBSTITUTE SHEET (RULE 26 CA 022~9314 1998-12-22 WO 98/00708 PCT/CAg7/00463 Where the likely se~uence of the DNA being analyzed is known, for example in repetitive diagnostic applications, the cost function used in the method of the invention may also compare the experimental data traces to a set of model data traces. In this case, a suitable cost function is the area between the experimental data trace and a corresponding data trace S from a set of model data traces that have been prealigned. The cost function is optimized when the area between the traces is at a minimum. Standard sets of model data traces comprised of random distributions of bases might also be used in such a comparison, although this increases the cost function space.
Another form of cost function that can be used in the method of the invention 10 evaluates the regularity of peak spacing achieved by the trial alignment. Such cost functions do not require normalization, because they do not depend on peak height or area. An example of such a function would be a function determining the standard deviation of the distance between adjacent peaks, excluding obvious outliers, with a low value for the cost function being desirable. Another example of a cost function of this type deterrnines the 15 distance between each peak and the closest base position within a postulated regular array of peak spacings. The cost function is optimized when the standard deviation of these distances for all of the peaks is at a minimum.
The postulated peak spacing in this array is tested over a several trial values established for a specific instrumental and experimental configuration, thus adding an 20 additional dimension to the cost function space. For example, where it is found that the normal peak spacing is 7 data points, trial peak spacings of 3 to 9 data points are suitably tested for each trial alignment to determine the lowest value of the cost function.
It will further be appreciated that the various cost functions discussed above can be used individually or in combination since they all work towards a common goal, the optimum 25 alignment of the data traces. Thus, alternative cost functions can be created as the product or dividend that results when two or more of the cost functions are combined.
Because the optimization of the stretch and shift coefficients is an iterative process involving the testing of many combination of trial values, when the cost space is too large reasonably to permit full-sampling of the cost space, it is desirable to use an optimization 30 routine which facilitates convergence towards an acceptable set of coefficients. It will be appreciated by persons skilled in the art that there are many types of optimization routines SVBSTITUTE SHEET (RULE 26) .. ... . ...

CA 022~9314 1998-12-22 using random or directed sparse sampling techniques which might be employed, including genetic a}gorithms and Monte Carlo techniques. A preferred method for determining the coefficients of stretch and shift that yield the optimum value of "cost" employs "simulated annealing."
Simulated annealing is a mathematical method of searching a broad parameter space for the "best fit" result, without having to test every member of the parameter space. This method is particularly relevant to signal alignment problems in high speed DNA sequencers because the parameter space for possible coefficient of stretch and shift needs to be extremely broad to accommodate the unpredictability and variation within each electrophoresis run.
The preterred manner of perforrning a simulated annealing calculation employs a computer. Many computer algorithms employing simulated annealing are known and available to those skilled in the art. Of particular interest are p~pers by Ingber et al.:Ingber, A.L., "Very fast simulated re-annealing," J Matlll. Comput. Modelling 12(8):
967-973 (1989);
Ingber, A.L. et al., "Genetic algorithms and very fast simulated re~nne~ling: A
comparison," JMathl. Comput. Modelling 16 (11): 87-100 (1992);
Ingber, A.L., "Simulated annealing: Practice versus theory," JMathl. Comput.
Modelling 18(11): 29-57 (1993); and Ingber, A.L., "Adaptive simulated annealing (ASA): Lessons learned, " J Control and Cvbernetics 25(1): 33-54 (1996).
Each of these papers is incorporated herein by reference.
As employed in the present invention, simulated annealing determines coefficients of stretch and shift for signal outputs from a Visible Genetics MicroGene BlasterTM as follows.
Consider a window of norrnalized signal output. The norrnalized data trace consists of a series of data points generated every 0.5 seconds. Each peak consists of about 6 to 7 data points, and requires 3 to 3.5 seconds to pass through the detection zone. For convenience of illustration, the window shown represents about 90 data points or about 15 peaks. In a prefelled embodiment, however, windows of 180 to 350 data points, representing about 30 to 50 peaks, most preferably of about 250 data points representing about 40 peaks are used.
One window is created for data from each of the four data traces, and the windows would be initially aligned on the basis of data-point number. Superimposing the four windows SUBSTITUTE SHEET (RULE 26~

CA 022~9314 1998-12-22 I I
reveals a non-minimi~ed "cost" result, that is, the amount of "cost" area is greater than it could be.
It is found that a first order equation can be applied to each point of a data trace within the window to modify its position and change the cost area:
SY=mX+b where Y = the new position of point X, b = offset (or shift), and m = stretch. A second order or higher equation (with coefficients in addition to stretch and shift being determined) is apparently not necessary to obtain satisfactory results, although second or higher order 10 equations may be used for more sophisticated analyses.
The parameter space for b and m is empirically selected. The offset coefficient (b) usually falls within 30 data points of the initial time-based alignment. Offset steps of 0.35 data points are suitably employed, thus providing a parameter space of 200 choices. For MicroGene ~31asterT~ data, this offset represents about S data peaks. The stretch coefficient 15 (m) usually falls within 5%. These steps are suitably made in 0.66% amounts. thus requiring 15 steps to cover the whole range. The total range of parameters for stretch and shift for each signal output is therefore 3000 (200 * 15). The range of parameters for alignment of three channels with respect to the fourth channel is therefore 7 billion (3000').
For each combination of the six parameters tested~ the coefficient are applied to the 20 three signal output functions and the functions are stretched and/or shifted accordingly. The adjusted functions are superimposed with the fourth signal output and the "cost" area is re-calculated. The cost will either be lower, higher or the same as before.
Efficient selection of the six parameters is crucial for the discovery of the parameters which provide the lowest cost. The simulated annealing theory selects parameters for testing 25 according to a variation of the Monte Carlo search technique "Boltzrnann Annealing" known as Adaptive Simulated Annealing (or Very Fast Simulated Re-Annealing). SimulatedAnnealing code is generally available to those skilled in the art over the Internet at http://www.ingber.com. The code provides operational steps for rapidly searching a large parameter space for an optimal solution given a cost function. An explanation of Simulated 30 Annealing is found in Ingber, A.L. "Very Fast Simulated Re-Annealing" J. Math. Comput.
Modeling ( 1989) 12: 967-973.

SUBSTITUTE SHEEr (RULE 26 CA 022~9314 1998-12-22 The simulated ~nne~ling technique employed in the invention uses algorithms which are well-known to those skilled in the mathematical arts in the following novel fashion.
Coefficients of stretch and shift are at first randomly selected and applied to the normalized data traces. The cost is calculated. New coefficients are then selected within a range defined by the annealing schedule (or temperature function "T") which governs the amount by which coefficients may be changed with each trial. The new coefficients are applied and cost is again determined. If the value of cost is lower than before, then the new point is used as the starting point for the next calculation. If the value of cost is higher, then the original coefficients are usually used again as the starting point of the next calculation. As in all simulated annealing processes, however, there is a finite probability P (initially on the order of about 20% or less) that the higher cost value will be used as the starting point. As the number of calculated values increase, T and P are reduced, thus tending to localize the search space around an area of low cost. Annealing temperature schedules allow the "temperature" parameter to be raised to a higher value again at intervals during the search, emulating the process of annealing used to heat treat metals. Eventually, when the search is fully completed, simulated annealing theory argues that the lowest cost value parameters will be found. In the above method, approximately 5000 sets of parameters are tested per calculation, representing 0.00001% of the available parameter field In practice~ the "fast annealing" modification of lngber is found to be satisfactory to obtain cost values low enough to base-call data from the Visible Genetics MicroGene BlasterTM (see Ingber, A.L. "Very Fast Simulated Re-Annealing" J. Mathl. Comput Modeling (1989) 12:967-973).
After the successful determination of the best fit parameter set for a given window of data points~ the next window of data is analyzed. The next window is selected to be the same number of data points as the first window, with an overlap of about 50% with the first window. Again, the coefficients of shift and stretch are identified which provide the lowest cost value when applied to the signal output functions. Thus, coefficienes are deterrnined for a series of piecewise domains, e.g., piecewise linear or cubic domains. The process of selecting windows and calculating coefficients continues until all the data has been analyzed.
When calculating the coefficients of stretch and shift for windows after the first window, it is sometimes advantageous to use the coefficients from a neighboring window as SUBSTITlJTE SHEET (RULE 26~

. .

CA 022~9314 1998-12-22 Wo 98/00708 PCT/CAg7/00463 the starting point for the simulated annealing process since the coefficients for neighboring windows tend to be related. In such cases, the annealing schedule T can be much shorter, for example testing only 1500 sets of coefficients as opposed to 5000. In addition, because it is desirable that the warp functions generated be continuous, subsequent windows may in fact be 5 evaluated as two "sub-windows." In the first sub-window, stretch coefficients are constrained such that the warp function does not change the offset already established at the center of the previous full window. In the second sub-window, the stretch is allowed to vary in a narrow range.
Ultimately, by this process of sliding a window in overlapping steps across the 10 combined norm~li7ed data, a "warp function" is arrived at for each norm~li7ed data trace.
This function reflects the relationship between optimum-cost shift values for each window. By connecting the determined points of the function, a curve is defined which gives coefficients of shift for each point in a data trace, and reflects stretch at each location by the slope of the curve.
As a general matter, warp functions are generated in this way for each data trace.
Thus, if the coefficients are determined with reference to a fixed standard, four different warp functions reflecting the varying ccefficients needed to align the four data traces are generated.
In practice, however. it will be understood that the warp functions may be determined relative to one of the four data traces. In this case, the coefficients for the one fixed data trace will all be one. In the context of this application, the phrase "generating a warp function for each norm~li7ecl data trace" encompasses both of these embodiments. The warp function may be represented by a plot of alignrnent of data points of the three warped traces, e.g., C, G and T
against A as shown in Fig. 8A or as a plot of displacement from the A trace versus data point as shown in Fig. 8B.
Each warp function is applied to its respective raw or conditioned data trace to adjust the alignment of the data trace and generate a "warped data trace." The four warped data traces are then combined in alignment to produce an aligned data set.
An additional peak spacing warp function may be generated and applied to adjust for variations in peak spacing as part of creating the warped data traces. The peak spacing warp function, and also the base separation function useable as a standard trace, are generated by SUBS~ITUTE SHEET (RULE 26~

.... . . .. . . .

CA 022~9314 1998-12-22 testing postulated peak separation values in successive windows of the data trace and minimi7ing a peak separation cost function for each window. A suitable cost function is COST=~, d-where d is the distance between each actual peak and a hypothetical peak located at the position fixed by the postulated peak separation value.
Presentation of the aligned data set may be done internally within a computer for use with base-calling functions. or it may involve display of the aligned data set on a video monitor. Either way, the presentation allows further use to be made of the modified output signals, for base-calling and other purposes. For example, the video display of aligned data may be useful to permit an operator to make manual adjustments, and to observe inaccuracies in base-calling.
Base-calling on the aligned data set may be performed in a variety of ways, including those base-calling techniques described in Tibbetts and Bowling (US Pat. No. 5,365,455) and Dam et al (US Pat. No. 5,1 19,316). Two preferred approaches to base-calling are described in detail below.
In the first approach for base-calling, peaks in each warped data trace making up the aligned data set are identified in the same manner in which peak detection was performed prior to norm~li7~tion of the data traces. A minimum peak height from the base-line may be selected by the opera~or to avoid spurious results. Identified peaks are then used for base-calling.
Occasionally, pealcs may represent a plurality of bands. It is necessary to deterrnine which peaks these are~ and how many bands they represent. An excellent method to employ is gaussian deconvolution whereby a peak is deconvolved into one or more standard g~lssi~n peaks representing singleton peaks. It is found that peaks generated from DNA sequencing reactions using T7 polyrnerase (Pharmacia, Sweden) and Thermo SequenaseTM (Amersham Life Sciences) generate the most consistent gaussian peaks.
The standard gaussian peak is deterrnined as shown in Fig. 9. Peaks are located in a conditioned data trace from one channel. A line is drawn between peak points. The point on the line halfway between peaks is joined to the data trace by a line L perpendicular to the baseline. The area under the curve A and between the two perpendicular lines (L(x), L(x+l)) SUBSTITUTE SHEE~ ULE 2~

CA 022~9314 1998-12-22 - lS -is deterrnined. Height (h) is measured from the baseline to the peak. h and A are used to calculate sigma (~J) according to the e4uation:

h~/~

where a represents the distance on the x-axis between the peak and the point at which the 5 value of the gaussian function v=e -(xla~-equals l/e.
For each detected peak, o is determined. For those peaks where the length of both L(x) and L(x+ I ) are greater than ~2 h, a linear regression is performed on the value of a.
10 Statistically, at most about 25% of peaks are expected to represent doubletons, triples or greater, so using a second linear regression to correct for a width trend over each window, the narrowest 50% of the peaks are selected for use in constructing a piecewise cubic "singleton width discriminant function" that specifies a model width of a singleton at each location in the data array. All peaks in the window that are narrower or equal to this function are deemed to 15 be singletons to a first approximation. This approximation may be further refined using constraints such as peak area, etc. (Fig. 10) In this way, a standard or model singleton gaussian peak height and width at any point on the data trace may be defined.
The characteristics of the standard g~llcsi~n peak(s) and the positions of the singletons found via the discriminant function in conjunction with the base separation function are then 20 used to classify all the peaks in the aligned data traces. The features (e.g., height, width and/or area) of the standard peak are compared to the features of a detected signal peak to determine the number of bases represented.
For example. if the standard spacing indicated by the base separation function is consistent with there being three peaks in a region between a pair of singletons and that region 25 is occupied by a large peak, the characteristics of the standard gaussian peak (area, height etc.) SUBSTITUTE SHEET (RULE 26) CA 022~9314 1998-12-22 are used to determine whether two, three or four peaks are most likely to be the number of base pairs represented by the large peak. Thus, for example, the difference between the area of the large peak and the area of the standard gaussian peak can be evaluated. If the area of the large peak is approximately three times the area of the standard peak, then the large peak is treated as representing three bases. Similarly, if the area is closer to four times the area of the standard peak, the large peak is treated as representing four bases. A similar, although less sensitive analysis can be performed based on the height of the peaks. The preferred analysis takes multiple peak features into account.
The primer peak may be eliminated or ignored as described for preparing the norm~li7ed signal output, so as not to interfere with base-calling.
Another method of identifying and elimin~ting the primer peak uses a peak counting method. The data stream is divided into windows of a certain number of data points. The peaks in each window are counted. When a primer peak is in the window, a window that normally would include 10 peaks, may have only 2 peaks. This window is eliminated from consideration. and other windows are used for alignment and base-calling.
Once the individual peaks are identified and the multiple peak curves are divided into individual gaussian peaks, the data may be base-called. Each peak is identified with one channel, representing a single base. Peaks are therefore assigned to specific bases. in sequential order, until the full sequence is identified.
Fig. I I outlines a second approach to base-calling using the aligned data set of the invention. In this case, peaks are identified within the aligned data set, for example as discussed above, and the data set is then divided into segments or windows. Within each window, a subset of the peaks are selected, for example 45% of the peaks, using selection criteria which select those peaks which are most likely to be "singleton peaks," i.e., peaks which represent only a single base and where that base is not the same as either adjacent base such that there should be no compression with adjacent peaks.
The selection of singleton peaks follows a several step process in which the features of each peak, i.e., width, height and area etc. are extracted and plotted in n-space. Based on the distribution of the peaks and the fact that, statistically, 56% of the bases in any seguence should be singletons within this definition, the region of n-space most likely to contain singleton peaks is defined. The peaks which fall within this region of space are then collected.

SUBSTITUTE SHEEr (RULE 26) CA 022~9314 1998-12-22 From the first pass singleton classification, piecewise singleton height and width functions that estimate model singleton features are defined for each segment or window.
These functions are then used to reclassify the peaks from the aligned data set as singleton or non-singleton peaks. From this refined singleton classification, a single piecewise single base separation distance function that estimates base-to-base separation is determined and used to provide postulated positions for all bases. These positions, combined with the actual peak features, neighboring characteristics and raw signal characteristics are then used to provide a best estimate of base call, together with a measure of the certainty of that call at each postulated base positions.
The method of the invention is advantageously practiced using a dedicated apparatus for determining nucleic acid sequences. As depicted in Fig. 12~ such an apparatus comprises a sequencer 90 having an electrophoresis gel holder 901 disposed between electrodes 902 which are used to apply an electric field to a gel placed in the holder to cause oligonucleotide fragments to migrate within the gel; a detection system comprising a source 908 for an interrogating beam 92 and a detector 907 for detecting the passage of oligonucleotide fragments through a detection zone. for example by monitoring emitted light 99; and a data processing system 96 operatively connected to the detector 95 for receiving raw data traces for each of the four chain termination product mixtures for a sample. Suitable gel holders, electrodes and detection systems are disclosed in US Patent Application No. 08/353.932, PCT
Patent Application No.PCTlUS95/15951, and US Patent Application No. US Patent Application No. 08/387,272, all of which are incorporated herein by reference~ although it will be understood that the particular configuration of the electrophoresis and detection system is not critical to the present invention.
The data processing system is suitably a personal or mini-computer which has stored therein a programmed instruction set effective to identify peaks in each of the four data traces;
normalize the height of said peaks in each of said data traces to a common value to generate four normalized data traces;
combine the four normalized data traces in an initial alignment;
determine coefficients of shift and stretch for selected data points within eachnormalized data trace, said coefficients optimi~ing a cost function which reflects the extent of SUBSTmJTE SHEET (F~UI E 26) CA 022~9314 1998-12-22 overlap of peaks in combined norrn~li7ed data traces to which the coefficients have been applied, said cost function being optimized when the extent of overlap is at a minimum;
generate a warp function for each norm:~li7e~1 data trace from the coefficients of shift and stretch determined for the selected data points;
apply each warp functions to the respective data trace or the norrn~li7ecl data trace to produce four warped data traces; and align the four warped data traces to form an aligned data set. The data processing system may be connected to a video display 97 for displaying the aligned data set.
[n a preferred embodiment of the invention~ the apparatus of the invention reports confidence levels to the system operator for some or all of the bases identified in the sequence.
The confidence level advantageously reflects both ( I ) the arithmetic agreement between the signal and the model. and (2) other features of the data signal (for example expanded peak width) which may indicate reasons that the confidence level should be lower than the apparent level based on arithmetic agreements. This confidence level can be reported for all peaks~ or it can be reported only for those pea!cs for which the confidence level falls below a selected threshold value. Peaks may also be flagged during the reporting process to report ambiguities in the identification of the number of bases represented by a multiple peak feature.

SUBSrlTUTE SHEEr (12ULE 26)

Claims (15)

1. A method for aligning data traces from four channels of an automated electrophoresis detection apparatus, each channel detecting the products of one of four chain-termination DNA sequencing reactions, whereby said four channels together provide information concerning the position of all four bases within a nucleic acid polymer being analyzed, comprising the steps of:
(a) identifying peaks in each of the four data traces;
(b) combining the four data traces in an initial alignment;
(c) determining coefficients of shift and stretch for selected data points within each data trace, said coefficients optimizing a cost function which reflects the suitability of a trial alignment resulting from the application of the trial stretch and shift coefficients to the combined data traces;
(d) generating a warp function for each data trace from the coefficients of shift and stretch determined for the selected data points;
(f) applying each warp functions to the respective data trace to produce four warped data traces; and (g) assembling the four warped data traces to form an aligned data set.
2. The method of claim 1, further comprising the step of normalizing the data traces to generate four normalized data traces in which homozygous peaks are all of substantially equal height prior to combining the four data traces into an initial alignment.
3. The method of claim 1 or 2, further comprising the step of displaying the aligned data set on a video display.
4. The method of any of claims 1 to 3, wherein the cost function reflects the extent of overlap of peaks in the combined normalized data traces to which the coefficients have been applied.
5. The method of any of claims 1 to 4, wherein the cost function determines the area of a region above the combined normalized data traces and below the common value, and wherein the cost function is optimized when this area is at a minimum.
6. The method of any of claims 1 to 4, wherein the cost function determines the area of a region below the combined normalized data traces, and wherein the cost function is optimized when this area is at a maximum.
7. The method of any of claims 1 to 3, wherein the cost function determines the area of a first region above the combined normalized data traces and the area of a second region, said second region being below the highest-edge of the combined normalized data traces and above the second highest edge of the combined data traces, wherein the cost function is optimized when the first area is minimized and the second area is maximized.
8. The method of any of claims 1 to 7, wherein the coefficients of shift and stretch for selected data points within each normalized data trace that yield the optimum value of the cost function are determined using a process of simulated annealing.
9. The method of any of claims 1 to 8, further comprising the step of conditioning the data traces prior to identification of the peaks in the data traces.
10. A method for determining the sequence of a nucleic acid polymer comprising the steps of:
(a) obtaining data traces from four channels of an automated electrophoresis detection apparatus, each channel detecting the products of one of four chain-termination DNA
sequencing reactions, whereby said four channels together provide information concerning the position of all four bases within a nucleic acid polymer being analyzed;
(b) aligning the data traces by a method according to any of claims 1 to 9 to produce an aligned data set;

(c) evaluating the aligned data set to determine the sequence of bases within the nucleic acid polymer.
11. The method according to claim 10, further comprising the steps of determining standard gaussian peak shapes for data points along the aligned data set;
assigning peaks in the aligned data set as singleton peaks or multiple peaks by comparison of the peaks in the aligned data set to the standard gaussian peak characteristics;
and determining how many standard gaussian peaks are contained within each multiple peak, wherein each multiple peak is treated as that number of singleton peaks for purposes of base-calling.
12. The method according to claim 10 or 11, further comprising the step of reporting confidence levels for at least of portion of the called base peaks.
13. The method according to claim 12, wherein the confidence level reflects the arithmetic agreement between the peak in the data trace and a model, and those features of the peak which might justify assignment of a lower confidence level.
14. An apparatus for determining the sequence of a nucleic acid polymer comprising:
(a) an electrophoresis gel holder;
(b) first and second electrodes disposed to apply an electric field to the electrophoresis gel disposed within the electrophoresis gel holder to cause oligonucleotide fragments loaded on the electrophoresis gel to migrate within the electrophoresis gel;
(c) a detection system comprising an interrogating beam and a detector for detecting the passage of oligonucleotide fragments through a detection zone; and(d) a data processing system operatively connected to the detector for receiving four data traces, one for each of four chain termination product mixtures for the nucleic acid polymer, wherein the data processing system has stored therein a programmed instruction set effective to identify peaks in each of the four data traces;
combine the four data traces in an initial alignment;
determine coefficients of shift and stretch for selected data points within eachdata trace, said coefficients optimizing a cost function which reflects the suitability of a trial alignment resulting from the application of the trial stretch and shift coefficients to the combined data traces;
generate a warp function for each data trace from the coefficients of shift and stretch determined for the selected data points;
apply each warp functions to the respective data trace to produce four warped data traces; and assemble the four warped data traces to form an aligned data set.
15. The apparatus according to claim 14, further comprising a video display for displaying the aligned data set.
CA002259314A 1996-06-27 1997-06-26 Method and apparatus for alignment of signals for use in dna base-calling Abandoned CA2259314A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US08/670,534 US5916747A (en) 1995-06-30 1996-06-27 Method and apparatus for alignment of signals for use in DNA based-calling
US08/670,534 1996-06-27

Publications (1)

Publication Number Publication Date
CA2259314A1 true CA2259314A1 (en) 1998-01-08

Family

ID=24690787

Family Applications (1)

Application Number Title Priority Date Filing Date
CA002259314A Abandoned CA2259314A1 (en) 1996-06-27 1997-06-26 Method and apparatus for alignment of signals for use in dna base-calling

Country Status (4)

Country Link
US (2) US5916747A (en)
AU (1) AU3250797A (en)
CA (1) CA2259314A1 (en)
WO (1) WO1998000708A1 (en)

Families Citing this family (63)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6090555A (en) * 1997-12-11 2000-07-18 Affymetrix, Inc. Scanned image alignment systems and methods
US6741344B1 (en) * 1994-02-10 2004-05-25 Affymetrix, Inc. Method and apparatus for detection of fluorescently labeled materials
US5631734A (en) 1994-02-10 1997-05-20 Affymetrix, Inc. Method and apparatus for detection of fluorescently labeled materials
US5853979A (en) 1995-06-30 1998-12-29 Visible Genetics Inc. Method and system for DNA sequence determination and mutation detection with reference to a standard
US5916747A (en) * 1995-06-30 1999-06-29 Visible Genetics Inc. Method and apparatus for alignment of signals for use in DNA based-calling
US6349144B1 (en) 1998-02-07 2002-02-19 Biodiscovery, Inc. Automated DNA array segmentation and analysis
US6990221B2 (en) * 1998-02-07 2006-01-24 Biodiscovery, Inc. Automated DNA array image segmentation and analysis
CA2328881A1 (en) * 1998-04-16 1999-10-21 Northeastern University Expert system for analysis of dna sequencing electropherograms
US6780591B2 (en) 1998-05-01 2004-08-24 Arizona Board Of Regents Method of determining the nucleotide sequence of oligonucleotides and DNA molecules
US7875440B2 (en) 1998-05-01 2011-01-25 Arizona Board Of Regents Method of determining the nucleotide sequence of oligonucleotides and DNA molecules
US6104027A (en) * 1998-06-05 2000-08-15 Hewlett-Packard Company Deconvolution of multiply charged ions
US6404907B1 (en) 1998-06-26 2002-06-11 Visible Genetics Inc. Method for sequencing nucleic acids with reduced errors
US6274317B1 (en) * 1998-11-02 2001-08-14 Millennium Pharmaceuticals, Inc. Automated allele caller
US6334099B1 (en) * 1999-05-25 2001-12-25 Digital Gene Technologies, Inc. Methods for normalization of experimental data
JP2003500662A (en) * 1999-05-25 2003-01-07 ディジタル・ジーン・テクノロジーズ・インコーポレーテッド Method and system for amplitude normalization and data peak selection
US7501245B2 (en) 1999-06-28 2009-03-10 Helicos Biosciences Corp. Methods and apparatuses for analyzing polynucleotide sequences
US7099502B2 (en) * 1999-10-12 2006-08-29 Biodiscovery, Inc. System and method for automatically processing microarrays
US6760668B1 (en) 2000-03-24 2004-07-06 Bayer Healthcare Llc Method for alignment of DNA sequences with enhanced accuracy and read length
US6436641B1 (en) * 2000-04-17 2002-08-20 Visible Genetics Inc. Method and apparatus for DNA sequencing
US20030170734A1 (en) * 2000-04-28 2003-09-11 Stephen Williams Multiplexed assays using electrophoretically separated molecular tags
US6474598B2 (en) * 2000-05-24 2002-11-05 Cartercopters, L.L.C. Landing gear shock absorber with variable viscosity fluid
US6591196B1 (en) * 2000-06-06 2003-07-08 Agilent Technologies Inc. Method and system for extracting data from surface array deposited features
DE60124598T2 (en) * 2000-06-06 2007-09-06 Agilent Technologies, Inc., Palo Alto Method and system for determining the orientation of a microarray
US6397150B1 (en) * 2000-07-27 2002-05-28 Visible Genetics Inc. Method and apparatus for sequencing of DNA using an internal calibrant
JP2004527728A (en) * 2000-08-14 2004-09-09 インサイト・ゲノミックス・インコーポレイテッド Base calling device and protocol
US6965704B2 (en) * 2000-08-22 2005-11-15 Affymetrix, Inc. System, method, and computer software product for grid alignment of multiple scanned images
US7062092B2 (en) 2000-08-22 2006-06-13 Affymetrix, Inc. System, method, and computer software product for gain adjustment in biological microarray scanner
EP1202049A1 (en) * 2000-10-31 2002-05-02 Genome Express Method and apparatus for analyzing data files derived from emission spectra from fluorophore tagged nucleotides
IL141151A0 (en) * 2001-01-29 2002-02-10 Compugen Ltd Method for comparing signal arrays in digital images
US6793790B1 (en) 2001-06-06 2004-09-21 The Regents Of The University Of California Sample collection system for gel electrophoresis
CA2453360A1 (en) * 2001-07-11 2003-01-23 Applera Corporation Internal calibration standards for electrophoretic analyses
AU2002329425A1 (en) * 2001-10-04 2003-04-14 Scientific Generics Limited Dna sequencer
US7222059B2 (en) * 2001-11-15 2007-05-22 Siemens Medical Solutions Diagnostics Electrophoretic trace simulator
WO2003085374A2 (en) * 2002-04-02 2003-10-16 Aclara Biosciences, Inc. Multiplexed assays using electrophoretically separated molecular tags
US20100266177A1 (en) * 2002-05-07 2010-10-21 Applied Biosystems, Llc Signal processing by iterative deconvolution of time series data
US20030228580A1 (en) * 2002-06-07 2003-12-11 Changsheng Liu Method for enhancing DNA base-calling accuracy
US7512496B2 (en) * 2002-09-25 2009-03-31 Soheil Shams Apparatus, method, and computer program product for determining confidence measures and combined confidence measures for assessing the quality of microarrays
WO2004029298A2 (en) * 2002-09-26 2004-04-08 Applera Corporation Mitochondrial dna autoscoring system
DE10315581B4 (en) * 2003-04-05 2007-06-28 Agilent Technologies, Inc. (n.d.Ges.d.Staates Delaware), Palo Alto Method for quality determination of RNA samples
US20040243341A1 (en) * 2003-05-30 2004-12-02 Le Cocq Christian A. Feature extraction methods and systems
US7169560B2 (en) 2003-11-12 2007-01-30 Helicos Biosciences Corporation Short cycle methods for sequencing polynucleotides
US20050115837A1 (en) * 2003-12-01 2005-06-02 Dean Burgi Analyte identification in transformed electropherograms
ATE463584T1 (en) 2004-02-19 2010-04-15 Helicos Biosciences Corp METHOD FOR ANALYZING POLYNUCLEOTIDE SEQUENCES
US20050281462A1 (en) * 2004-06-16 2005-12-22 Jayati Ghosh System and method of automated processing of multiple microarray images
US7647188B2 (en) * 2004-09-15 2010-01-12 F. Hoffmann-La Roche Ag Systems and methods for processing nucleic acid chromatograms
US7877213B2 (en) * 2004-09-23 2011-01-25 Agilent Technologies, Inc. System and methods for automated processing of multiple chemical arrays
US20060160102A1 (en) 2005-01-18 2006-07-20 Hossein Fakhrai-Rad Identification of rare alleles by enzymatic enrichment of mismatched heteroduplexes
US7666593B2 (en) 2005-08-26 2010-02-23 Helicos Biosciences Corporation Single molecule sequencing of captured nucleic acids
GB0522335D0 (en) * 2005-11-02 2005-12-07 Deltadot Ltd Improvements in sequencing
JP4922646B2 (en) * 2005-12-14 2012-04-25 株式会社日立ソリューションズ Gene information display method and display device
US9445025B2 (en) 2006-01-27 2016-09-13 Affymetrix, Inc. System, method, and product for imaging probe arrays with small feature sizes
US8055098B2 (en) 2006-01-27 2011-11-08 Affymetrix, Inc. System, method, and product for imaging probe arrays with small feature sizes
WO2007092849A2 (en) * 2006-02-06 2007-08-16 Siemens Healthcare Diagnostics Inc. Methods for detecting peaks in a nucleic acid data trace
US20130090860A1 (en) * 2010-12-30 2013-04-11 Life Technologies Corporation Methods, systems, and computer readable media for making base calls in nucleic acid sequencing
US10689696B2 (en) 2013-12-03 2020-06-23 Illumina, Inc. Methods and systems for analyzing image data
CN103760202B (en) * 2014-01-28 2016-03-09 东北师范大学 A kind of method using the electrochemical sensor of preparation simultaneously to detect DNA base
US10217379B2 (en) 2015-01-30 2019-02-26 Toyota Motor Engineering & Manufacturing North America, Inc. Modifying vision-assist device parameters based on an environment classification
US9914218B2 (en) 2015-01-30 2018-03-13 Toyota Motor Engineering & Manufacturing North America, Inc. Methods and apparatuses for responding to a detected event by a robot
US10037712B2 (en) 2015-01-30 2018-07-31 Toyota Motor Engineering & Manufacturing North America, Inc. Vision-assist devices and methods of detecting a classification of an object
US9508165B1 (en) * 2015-06-30 2016-11-29 General Electric Company Systems and methods for peak tracking and gain adjustment
JP7230208B2 (en) * 2018-12-10 2023-02-28 ライフ テクノロジーズ コーポレーション Sanger Sequencing Deep Bass Cola
CN110263708B (en) * 2019-06-19 2020-03-13 郭玮强 Image source identification method, device and computer readable storage medium
CN113658113B (en) * 2021-07-28 2024-02-27 武汉联影医疗科技有限公司 Medical image detection method and training method of medical image detection model

Family Cites Families (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE2056998A1 (en) * 1970-11-20 1972-05-25 Bodenseewerk Perkin Elmer Co Peak integrator
JPS55164336A (en) * 1979-06-08 1980-12-22 Olympus Optical Co Ltd Inspecting material detecting method
US5360523A (en) * 1984-03-29 1994-11-01 Li-Cor, Inc. DNA sequencing
EP0198403A3 (en) * 1985-04-09 1989-07-19 Fuji Photo Film Co., Ltd. Signal processing method for determining base sequence of nucleic acid
US4885696A (en) * 1986-03-26 1989-12-05 Fuji Photo Film Co., Ltd. Signal processing method for determining base sequence of nucleic acid
US4811218A (en) * 1986-06-02 1989-03-07 Applied Biosystems, Inc. Real time scanning electrophoresis apparatus for DNA sequencing
US4823007A (en) * 1986-12-15 1989-04-18 Norand Corporation DNA sequencing gel reading system and method
GB8826816D0 (en) * 1988-11-16 1988-12-21 Atomic Energy Authority Uk Method for spectrum matching
US5119316A (en) * 1990-06-29 1992-06-02 E. I. Du Pont De Nemours And Company Method for determining dna sequences
US5419825A (en) * 1991-07-29 1995-05-30 Shimadzu Corporation Base sequencing apparatus
US5502773A (en) * 1991-09-20 1996-03-26 Vanderbilt University Method and apparatus for automated processing of DNA sequence data
US5365455A (en) * 1991-09-20 1994-11-15 Vanderbilt University Method and apparatus for automatic nucleic acid sequence determination
US5308751A (en) 1992-03-23 1994-05-03 General Atomics Method for sequencing double-stranded DNA
JP2815506B2 (en) * 1992-04-14 1998-10-27 株式会社日立製作所 Light detection type electrophoresis device
EP0592060B1 (en) * 1992-10-09 2001-07-18 The Board of Regents of the University of Nebraska Digital DNA typing
US5273632A (en) * 1992-11-19 1993-12-28 University Of Utah Research Foundation Methods and apparatus for analysis of chromatographic migration patterns
US5666435A (en) * 1994-12-09 1997-09-09 Genomyx Corporation System for analysis of x-ray films of nucleotide sequences
US5916747A (en) * 1995-06-30 1999-06-29 Visible Genetics Inc. Method and apparatus for alignment of signals for use in DNA based-calling
EP0896632B1 (en) 1996-05-01 2001-12-12 Visible Genetics Inc. Method for detection and identification of microorganisms
US6208941B1 (en) 1996-09-16 2001-03-27 University Of Utah Research Foundation Method and apparatus for analysis of chromatographic migration patterns

Also Published As

Publication number Publication date
WO1998000708A1 (en) 1998-01-08
US6554987B1 (en) 2003-04-29
US5916747A (en) 1999-06-29
AU3250797A (en) 1998-01-21

Similar Documents

Publication Publication Date Title
US6554987B1 (en) Method and apparatus for alignment of signals for use in DNA base-calling
US5853979A (en) Method and system for DNA sequence determination and mutation detection with reference to a standard
US10991453B2 (en) Alignment of nucleic acid sequences containing homopolymers based on signal values measured for nucleotide incorporations
US6015667A (en) Multicomponent analysis method including the determination of a statistical confidence interval
Frith et al. Detection of cis-element clusters in higher eukaryotic DNA
US7228237B2 (en) Automatic threshold setting and baseline determination for real-time PCR
US7406385B2 (en) System and method for consensus-calling with per-base quality values for sample assemblies
US20140051154A1 (en) Systems and methods to detect copy number variation
WO1998008085A1 (en) Automatic sequencer/genotyper having extended spectral response
US20080306694A1 (en) Methods for Detecting Peaks in a Nucleic Acid Data Trace
US20100280758A1 (en) method and an apparatus for determining nucleotide sequence, and a computer program product to be executed by the apparatus
US7799527B2 (en) Systems and methods for automated quality control of polymer synthesis
US5891632A (en) Method and device for fragment trace data display in DNA base sequencing
US6203990B1 (en) Method and system for pattern analysis, such as for analyzing oligonucleotide primer extension assay products
JP2004527728A (en) Base calling device and protocol
CN107991411A (en) It is used for the method for mass spectrum biopolymer analysis using the oligomer scheduling of optimization
CN116134526A (en) DNA analyzer with synthetic allele ladder library
JPH05168500A (en) Determination of nucleic acid base sequence
JP2875849B2 (en) Method and apparatus for determining nucleic acid base sequence
Kramer Overview of the Tools for Microarray Analysis: Transcription Profiling, DNA Chips, and Differential Display
EP1202049A1 (en) Method and apparatus for analyzing data files derived from emission spectra from fluorophore tagged nucleotides

Legal Events

Date Code Title Description
EEER Examination request
FZDE Discontinued