Methods and apparatus for analysis of chromatographic migration patterns

• US 5,273,632 A
• Filed: 11/19/1992
• Issued: 12/28/1993
• Est. Priority Date: 11/19/1992
• Status: Expired due to Term

• Current Assignee: University of Utah Research Foundation (State of Utah)
• Original Assignee: University of Utah Research Foundation (State of Utah)
• Inventors: Ives, Jeffrey T., Stockham, Thomas G.
• Primary Examiner(s): Niebling, John
• Assistant Examiner(s): DELACROIX MUIRHEI, CYBILLE
• Application Number:
  US07/978,915
• Time in Patent Office: 404 Days
• Field of Search: 204/299 R, 204/182.8, 204/180.1, 364/497, 210/656, 935/77, 935/85, 935/86, 935/87
• US Class Current: 204/450
• CPC Class Codes: G01N 27/44717 Arrangements for investigating the separated zones, e.g. localising zones
• CPC Class Codes: G01N 27/44721 by optical means
• CPC Class Codes: G01N 30/8624 Detection of slopes or peaks; baseline correction
• CPC Class Codes: G01N 30/8631 Peaks
• CPC Class Codes: Y10T 436/143333 Saccharide [e.g., DNA, etc.]

Abstract

A method and apparatus for sharpening signal peaks in a signal representing the distribution of biological or chemical components of a mixture separated by a chromatographic technique such as, but not limited to, electrophoresis. A key step in the method is the use of a blind deconvolution technique, presently embodied as homomorphic filtering, to reduce the contribution of a blurring function to the signal encoding the peaks of the distribution. The invention further includes steps and apparatus directed to determination of a nucleotide sequence from a set of four such signals representing DNA sequence data derived by electrophoretic means.

Claims

1. A method of sharpening peaks in a signal representing a chromatographic distribution of components from a biochemical mixture separated by a chromatographic method comprising the steps of:

providing a signal representing a chromatographic distribution of components from a biochemical mixture separated by a chromatographic method;

transforming the signal from its original space domain to a cepstrum;

manipulating the cepstrum with a lifter function selected to substantially reduce the amplitude of a portion of the cepstrum which is attributable to a blurring function, thereby producing a liftered cepstrum signal; and

de-transforming the liftered cepstrum signal to produce a deconvolved lane signal in the original space domain.

2. The method of claim 1, further including a step of noise filtering for substantially removing the effects of background noise in the deconvolved lane signal.

3. The method of claim 2, wherein said step of noise filtering is performed in conjunction with said step of de-transforming.

4. The method of claim 3, wherein said step of noise filtering comprises applying a Gaussian lowpass filter having a bandwidth of between about 0.024A and about 0.072A, where A=the number of samples comprising π

.

5. The method of claim 2, wherein in said step of manipulating the cepstrum, the cepstrum is multiplied by the lifter function and the lifter function is configured to have a first portion which attenuates the cepstrum in a low-frequency region of the cepstrum and a second portion which is substantially equal to one in a high-quefrency region of the cepstrum.

6. The method of claim 5, wherein the first portion of the lifter function has the approximate shape of a 50% cosine taper which reaches an ordinate value of 0.50 at a selected point between the low-quefrency region and the high-quefrency region of the cepstrum.

7. The method of claim 6, wherein said selected point is located approximately in the region in which the value of the cepstrum reaches a plateau.

8. A method of sharpening peaks in a signal representing a chromatographic distribution of components from a biochemical mixture separated by a chromatographic method comprising the steps of:

providing a signal representing a chromatographic distribution of components from a biochemical mixture separated by a chromatographic method;

generating an FS (frequency spectrum) signal by Fourier transformation of the signal;

taking the log of the FS signal to produce a CLS (complex log-spectrum) signal;

taking an inverse Fourier transform of the real portion of the CLS signal to produce a cepstrum signal;

multiplying the cepstrum signal by a lifter function chosen to reduce the contribution of a blurring function, thereby producing a liftered cepstrum signal;

subjecting the liftered cepstrum signal to Fourier transformation to produce an LLS (liftered log-spectrum) signal;

adding the imaginary portion of the LS signal to the LLS signal to produce a liftered CLS signal;

taking the inverse logarithm of the liftered CLS signal to produce a liftered FS (liftered frequency spectrum) signal; and

taking the inverse Fourier transform of the liftered FS signal to produce a deconvolved signal.

9. The method of claim 8, further including a step of removing noise from the signal being processed, said step of removing noise being performed after said step of producing an FS signal.

10. The method of claim 9, wherein said step of removing noise is performed upon said liftered FS signal before said step of taking the inverse Fourier transform, and comprises applying a low-pass filter to said liftered FS signal, to substantially remove frequencies above a selected cut-off frequency and produce a noise-filtered FS signal.

11. The method of claim 10, wherein said low-pass filter is a Gaussian filter having a bandwidth equivalent to between about 50 and 150 frequency samples when π

is 1024 samples.

12. The method of claim 9, wherein the lifter function is a high-pass type having a first portion which attenuates the cepstrum in a low-frequency region of the cepstrum and a second portion which is substantially constant through a high-frequency region of the cepstrum

13. The method of claim 12 wherein the second portion of the lifter function is selected to normalize the amplitude of the cepstrum to a chosen range.

14. The method of claim 13, wherein the first portion of the lifter function approximately conforms to a 50% cosine taper which reaches an ordinate value of 0.50 at a selected point between the low-frequency region and the high-frequency region of the cepstrum.

15. The method of claim 13, wherein said selected point is located approximately in the region in which the value of the cepstrum reaches a plateau.

16. The method of claim 9, wherein the biochemical mixture is selected from the group consisting of:

mixtures of polypeptides, sets of DNA fragments generated in a DNA sequencing reaction, mixtures of organic chemicals, and mixtures of fluorescently labelled cells or subcellular components including organelles, nuclei, chromosomes, and fragments thereof.

17. The method of claim 16, wherein the chromatographic method is selected from the group consisting of:

electrophoresis, high-pressure liquid chromatography, fluorescence-activated separation, affinity chromatography, thin-layer chromatography, paper chromatography.

18. A method of determining a nucleotide sequence of a DNA molecule from an electrophoretic migration pattern of a set of DNA sequencing lanes sufficient to establish the relative migration patterns of fragment groups respectively terminating in each one of the nucleotides designated A, T, G and C, comprising the steps of:

providing a set of lane signals respectively encoding the migration patterns of each member of the set of sequencing lanes, and each lane signal having peaks representing the relative amounts of DNA fragments of different sizes present in the corresponding lane;

identifying putative peaks in all of the lane signals and determining a putative spacing which is the average for all of the putative peaks;

aligning the lane signals to establish a provisional ordering of the putative peaks;

generating a three-dimensional matrix having a plurality of matrix elements, each matrix element taking the form of a coordinate pair comprising a spacing value for a specific occurrence of a peak pair, and the location in the signal of that specific occurrence, wherein a peak pair is defined as a pair of non-identical peaks which are adjacent each other in the provisional ordering;

deriving a peak pair spacing function for each category of peak pair;

selecting a reference lane; and

adjusting the number of samples within each lane signal as needed to produce a spacing function for the lane signal which substantially matches the spacing function of the reference lane, thereby producing a set of fully aligned lane signals; and

reading a DNA sequence from the order of peaks in the aligned lane signals

19. The method of claim 18, further including a step of fitting the peak pair spacing function to a straight line, performed prior to said step of selecting a reference lane.

20. The method of claim 18, wherein said step of identifying putative peaks comprises the steps of establishing a threshold function and selecting peaks whose intensity exceeds the value of the threshold function.

21. The method of claim 20, wherein the threshold function is a variable threshold function selected to reflect the variation in average peak intensity with position in the signal.

22. The method of claim 21, further including a step of coarse alignment of the lane signals, said step of coarse alignment being performed after said step of identifying putative peaks and before said step of generating a three dimensional matrix.

23. The method of claim 22, wherein said step of coarse alignment includes the following substeps:

deriving a plurality of peak pair spacings, one corresponding to each possible non-identical peak pair;

comparing each of the peak pair spacings to the putative spacing;

selecting a coarse reference lane; and

shifting by a selected increment relative to the coarse reference lane, each non-reference lane for which the absolute value of the difference between its individual peak pair spacing and the putative spacing differs by a significant amount from the value for the coarse reference lane.

24. The method of claim 23, wherein after said step of shifting by an increment, the following further steps are performed, comprising:

re-computing the peak pair spacing for the shifted lanes to produce a corresponding number of adjusted peak pair spacings and comparing the absolute difference value for each shifted lane to that of the coarse reference lane; and

interatively repeating said steps of shifting lanes, re-computing the peak pair spacing, and comparing the absolute difference values until a satisfactory coarse alignment is achieved.

25. The method of claim 18, further including a step of blind deconvolution of said lane signals performed prior to said step of identifying putative peaks.

26. The method of claim 25, wherein said step of blind deconvolution comprises the following steps:

transforming the signal from its original space domain to a cepstrum;

manipulating the cepstrum with a lifter function selected to substantially reduce the amplitude of a portion of the cepstrum which is attributable to a blurring function, thereby producing a liftered cepstrum signal; and

de-transforming the liftered cepstrum signal to produce a deconvolved lane signal in the original space domain.

27. The method of claim 26, further including a step of noise filtering performed in conjunction with said step of de-transforming to produce a noise-filtered deconvolved signal.

28. The method of claim 27, wherein in said step of manipulating the cepstrum, the cepstrum is multiplied by the lifter function and the lifter function is configured to have a first portion which attenuates the cepstrum in a low-frequency region of the cepstrum and a second portion which is substantially equal to one in a high-frequency region of the cepstrum.

29. A method of aligning the members of a set of lanes of a DNA sequencing electrophorogram, comprising the steps of:

providing a set of lane signals respectively encoding the migration patterns of each member of the set of sequencing lanes, and each lane signal having peaks representing the relative amounts of DNA fragments of different sizes present in the corresponding lane;

establishing a provisional alignment of the lanes;

generating a three-dimensional matrix having matrix elements in the form of coordinate pairs, each coordinate pair comprising a peak pair spacing value for an individual occurrence of that peak pair, and the signal position of the individual occurrence;

deriving a peak pair spacing function for each category of peak pair, and fitting the spacing function to a straight line;

selecting a reference lane; and

adjusting the number of samples within each lane signal as needed to produce a spacing function for the lane signal which substantially matches the spacing function of the reference lane.

30. The method of claim 29, wherein said step of selecting a reference lane is performed by selecting the lane whose spacing function has the largest slope value when compared with the spacing functions of the other members of the set.

31. The method of claim 30, wherein said step of establishing a provisional alignment comprises the steps of:

aligning the lane signals to establish a provisional ordering of the putative peaks;

deriving a plurality of peak pair spacings, one corresponding to each possible non-identical peak pair;

comparing each of the peak pair spacings to the putative spacing;

selecting a coarse reference lane;

shifting by a selected increment relative to the coarse reference lane, each non-reference lane for which the absolute value of the difference between its individual peak pair spacing and the putative spacing differs by a significant amount from the value for the coarse reference lane;

re-computing the peak pair spacing for the shifted lanes to produce a corresponding number of adjusted peak pair spacings and comparing the absolute difference value for each shifted lane to that of the coarse reference lane; and

interatively repeating said steps of shifting lanes, re-computing the peak pair spacing, and comparing the absolute difference values until a satisfactory coarse alignment is achieved.

32. The method of claim 30, wherein said step of establishing a provisional alignment includes the steps of:

identifying putative peaks by determining a variable threshold function based on the variation in average peak intensity with position in the signal; and

selecting peaks whose intensity exceeds the value of the threshold function.

33. A method of determining a DNA sequence from an electrophorogram of a set of nucleic acid sequencing lanes, comprising the steps of:

providing a set of lane signals, each representing a corresponding one of a set of sequencing lanes and comprising a peak signal function convolved with a blurring function, the peak signal function comprising peaks reflective of the relative amounts of particular fragments in the sequencing lane;

processing by blind deconvolution each of said lane signals to reduce the contribution of the blurring function to each lane signal, thereby producing a set of corresponding deconvolved lane signals;

mutually aligning said lane signals to establish an ordering of peaks among the respective lanes; and

reading a nucleic acid sequence from the order of peaks in the aligned lane signals.