Method for improving the sensitivity of detection in determining copy number variations

• US 10,741,269 B2
• Filed: 10/21/2014
• Issued: 08/11/2020
• Est. Priority Date: 10/21/2013
• Status: Active Grant

• Current Assignee: Verinata Health Incorporated (Illumina Incorporated)
• Original Assignee: Verinata Health Incorporated (Illumina Incorporated)
• Inventors: Chudova, Darya I., Abdueva, Diana, Rava, Richard P.
• Primary Examiner(s): Leith, Nancy J
• Application Number:
  US15/031,246
• Publication Number:
  US 20160239604A1
• Time in Patent Office: 2,121 Days
• Field of Search: None
• CPC Class Codes: G16B 20/10 Ploidy or copy number detection
• CPC Class Codes: G16B 25/00 ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression
• CPC Class Codes: G16B 25/10 Gene or protein expression profiling; Expression-ratio estimation or normalisation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Disclosed are methods for determining copy number variation (CNV) known or suspected to be associated with a variety of medical conditions. In some embodiments, methods are provided for determining copy number variation (CNV) of fetuses using maternal samples comprising maternal and fetal cell free DNA. In some embodiments, methods are provided for determining CNVs known or suspected to be associated with a variety of medical conditions. Some embodiments disclosed herein provide methods to improve the sensitivity and/or specificity of sequence data analysis by removing within-sample GC-content bias. In some embodiments, removal of within-sample GC-content bias is based on sequence data corrected for systematic variation common across unaffected training samples. Also disclosed are systems and computer program products for evaluation of CNV of sequences of interest.

Claims

1. A method, implemented at a computer system that includes one or more processors and system memory, for evaluation of copy number of a nucleic acid sequence of interest in a test sample, the method comprising:

(a) providing, at the computer system, at least 10,000 sequence reads obtained by a nucleic acid sequencer from the test sample, which test sample comprises nucleic acid molecules from one or more genomes;

(b) aligning, by the computer system, the at least 10,000 sequence reads of the test sample to a reference genome comprising the nucleic acid sequence of interest, thereby providing test sequence tags;

(c) determining, by the computer system, a coverage of the test sequence tags located in each bin, wherein each chromosome of the reference genome is divided into a plurality of bins, and wherein the coverage indicates a quantity of sequence tags in a bin;

(d) providing, by the computer system, a global profile for the nucleic acid sequence of interest, wherein the global profile comprises an expected coverage in each bin, and wherein the expected coverage is obtained from a training set of training samples unaffected by a copy number variation of the nucleic acid sequence of interest, the expected coverage exhibiting variation from bin to bin;

(e) adjusting, by the computer system, the coverage of the test sequence tags in each bin of at least the nucleic acid sequence of interest using the expected coverage in each bin, thereby obtaining global-profile-corrected coverages for the nucleic acid sequence of interest;

(f) adjusting, by the computer system, the global-profile-corrected coverages based on a relation between GC content levels of the test sample and the global-profile-corrected coverages of the test sample, thereby obtaining sample-GC-corrected coverages for the nucleic acid sequence of interest, wherein the adjusting is not based on GC-coverage relations of samples other than the test sample; and

(g) evaluating, by the computer system, a copy number of the nucleic acid sequence of interest in the test sample based on the sample-GC-corrected coverages, wherein the sample-GC-corrected coverages improve a signal level and/or reduce a noise level for determining the copy number of the nucleic acid sequence of interest.

2. The method of claim 1, further comprising, before providing sequence reads, sequencing nucleic acids from the test sample using a sequencer, thereby generating the sequence reads.

3. The method of claim 2, further comprising combining marker nucleic acids with the test sample before sequencing the nucleic acids.

4. The method of claim 3, wherein the marker nucleic acids are selected from the group consisting of:

naturally-occurring deoxyribonucleic acids, naturally-occurring ribonucleic acids, peptide nucleic acids (PNA), morpholino nucleic acid, locked nucleic acids, glycol nucleic acids, threose nucleic acids, and any combinations thereof.

5. The method of claim 1, wherein the sequence reads are obtained from sequences of cell-free DNA of a pregnant woman and cell-free DNA of a fetus carried by the pregnant woman.

6. The method of claim 1, further comprising applying a sequence mask that excludes from consideration coverages in masked bins.

7. The method of claim 6, wherein the sequence mask is obtained by a method comprising:

providing, at the computer system, a training set comprising sequence reads from a plurality of unaffected training samples;

aligning, by the computer system, the sequence reads of the training set to the reference genome, thereby providing training sequence tags for the training samples;

dividing, by the computer system, the reference genome into a plurality of bins;

determining, by the computer system, a coverage of training sequence tags in each bin for each training sample; and

creating, by the computer system, a sequence mask comprising unmasked and masked bins, wherein each masked bin has a distribution index exceeding a masking threshold, the distribution index relating to the distribution of coverages of the training samples.

8. The method of claim 7, further comprising, before creating a sequence mask, adjusting the coverage of the training sequence tags according to the expected coverage in each bin, thereby obtaining global-profile-corrected coverages of the training sequence tags in the bins, which are then used to create a sequence mask.

9. The method of claim 7, wherein the distribution index is mathematically related to the variance of the coverages of the training samples.

10. The method of claim 9, wherein the distribution index is a coefficient of variation.

11. The method of claim 6, wherein the masked bins on the nucleic acid sequence of interest have a first masking threshold and the masked bins on a normalizing sequence have a second masking threshold.

12. The method of claim 11, wherein the combination of the first masking threshold and the second masking threshold provides a sequence mask that results in a lower variation of coverage over a region including the sequence of interest in unaffected samples than sequence masks obtained using other thresholds.

13. The method of claim 6, wherein the sequence mask comprises masked bins and unmasked bins defined by the distribution of mapping quality scores across training samples within the bins, the mapping quality scores being derived from aligning sequence reads of a plurality of unaffected training samples to the reference genome.

14. The method of claim 1, wherein evaluating copy number of the nucleic acid sequence of interest in the test sample in operation (g) comprises calculating a sequence dose of the nucleic acid sequence of interest for the test sample using coverage information of a normalizing sequence.

15. The method of claim 14, wherein calculating the sequence dose comprises dividing the sample-GC-corrected coverage of the test sequence tags in the nucleic acid sequence of interest by the sample-GC-corrected coverage of the test sequence tags in a normalizing sequence.

16. The method claim 15, wherein the normalizing sequence comprises one or more robust autosomes sequences or segments thereof.

17. The method of claim 1, wherein evaluating the copy number of the nucleic acid sequence of interest in the test sample in operation (g) comprises calculating a normalized chromosome value or a normalized segment value of the nucleic acid sequence of interest for the test sample using coverage information of a normalizing sequence.

18. The method of claim 1, wherein the test sample comprises a mixture of nucleic acids from two different genomes.

19. The method of claim 18, wherein said nucleic acids comprise cell-free DNA molecules.

20. The method of claim 1, wherein the test sample comprises fetal and maternal cell-free nucleic acids.

21. The method of claim 1, wherein the test sample comprises fetal cell-free nucleic acids from two or more fetuses.

22. The method of claim 1, wherein the test sample comprises nucleic acids from cancerous and unaffected cells from the same subject.

23. The method of claim 1, wherein the evaluating copy number of the nucleic acid sequence of interest in the test sample comprises determining the presence or absence of a complete or partial fetal aneuploidy.

24. The method of claim 1, further comprising removing, after operation (f), outlier bins of sample-GC-corrected coverage from consideration in evaluation of copy number variation.

25. The method of claim 24, wherein the outlier bins comprises bins whose median sample-GC-corrected coverage are more than about 1 median absolute deviation from the median of all the bins.

26. The method of claim 1, wherein the expected coverage in each bin comprises a median or mean of coverages of training samples, and wherein the adjusting the coverage of the test sequence tags in operation (e) comprises dividing the coverage of the test sequence tags for each bin by the median or mean of coverages of training samples from the bin.

27. The method of claim 1, wherein adjusting the coverage of the test sequence tags in operation (e) comprises:

(i) obtaining a relation between the coverage of the test sequence tags versus the expected coverage in a plurality of bins in one or more robust chromosomes or regions, and (ii) applying the relation to bins in the sequence of interest to obtain the global-profile-corrected coverages.

28. The method claim 27, wherein

the relation in (i) is a obtained by a linear regression:

y_a=intercept+slope*gwp_a,

wherein y_a is coverage of bin a for the test sample in one or more robust chromosomes or regions, and gwp_a is the global profile for bin a for unaffected training samples; and

obtaining the global-profile-corrected coverages in (ii) comprises obtaining the global-profile-corrected coverages z_b as;

z_b=y_b/(intercept+slope*gwp_b)−

1

wherein y_b is an observed coverage of bin b for the test sample in the sequence of interest, and gwp_b is the global profile for bin b for unaffected training samples.

29. The method of claim 1, wherein the global-profile-corrected coverage of the test sequence tags from (e) comprise global-profile-corrected coverage of bins in the nucleic acid sequence of interest and global-profile-corrected coverage of bins in a normalizing sequence.

30. The method of claim 1, wherein adjusting the global-profile-corrected coverage in operation (f) comprises:

grouping bins in the reference genome into a plurality of GC groups, each GC group comprising multiple bins, wherein the multiple bins contain test sequence tags and have similar GC content;

determining an expected value of the global-profile-corrected coverage for each GC group for a plurality of robust autosomes; and

adjusting the global-profile-corrected coverage of the test sequence tags for each GC group based on the determined expected value of the same GC group, thereby obtaining the sample-GC-corrected coverage of the test sequence tags on the nucleic acid sequence of interest.

31. The method of claim 30, wherein the expected value of the global-profile-corrected coverage is the mean or the median of the coverage for the GC group of a plurality of robust autosomes.

32. The method of claim 30, wherein adjusting the global-profile-corrected coverage of the test sequence tags comprises subtracting the expected value from the global-profile-corrected coverage.

33. The method of claim 30, wherein the adjusting the global-profile-corrected coverage in operation (f) comprises:

fitting a linear or non-linear mathematical function to data points from a plurality of robust autosomes, wherein each data point relates a coverage value to a GC content value; and

adjusting the global-profile-corrected coverage of the test sequence tags in each bin based on an expected value of coverage for each bin, which equals to the coverage value of the mathematical function at the GC content value of the bin under consideration.

34. The method of claim 33, wherein adjusting the global-profile-corrected coverage of the test sequence tags comprises subtracting the expected value from the global-profile-corrected coverage.

35. The method of claim 1, further comprising extracting cell free DNA from a plurality of unaffected individuals and/or the test sample.

36. The method of claim 1, wherein the sequence reads comprise sequences of about 20 to 50-bp from anywhere in the entire genome of an individual.

37. The method of claim 1, wherein the sequence reads of (a) comprises bar-coded 25-mers.

38. The method of claim 1, wherein the sample is a blood plasma sample.

39. A system for evaluation of copy number of a nucleic acid sequence of interest in a test sample, the system comprising:

a sequencer for receiving nucleic acids from the test sample providing nucleic acid sequence information from the sample;

one or more processors; and

one or more computer-readable storage media having stored thereon instructions for execution on said processor to evaluate copy number in the test sample using a method comprising;

(a) providing, at the system, at least 10,000 sequence reads of the test sample;

(b) aligning, by the one or more processors, the at least 10,000 sequence reads of the test sample to a reference genome comprising the nucleic acid sequence of interest, thereby providing test sequence tags;

(c) determining, by the one or more processors, a coverage of the test sequence tags located in each bin, wherein each chromosome of the reference genome is divided into a plurality of bins;

(d) providing, by the one or more processors, a global profile for the nucleic acid sequence of interest, wherein the global profile comprises an expected coverage in each bin, and wherein the expected coverage is obtained from a training set of training samples unaffected by a copy number variation of the nucleic acid sequence of interest, the expected coverage exhibiting variation from bin to bin;

(e) adjusting, by the one or more processors, the coverage of the test sequence tags in each bin of at least the nucleic acid sequence of interest according to the expected coverage in each bin, thereby obtaining a global-profile-corrected coverage in each bin of the test sequence tags;

(f) adjusting, by the one or more processors, the global-profile-corrected coverages based on a relation between GC content level of the test sample and the global-profile-corrected coverage of the test sample for the bins of the test sequence tags, thereby obtaining a sample-GC-corrected coverage of the test sequence tags on the nucleic acid sequence of interest, wherein the adjusting is not based on GC-coverage relations of samples other than the test sample; and

(g) evaluating, by the one or more processors, copy number of the nucleic acid sequence of interest in the test sample based on the sample-GC-corrected coverage.