Methods and processes for non-invasive detection of a microduplication or a microdeletion with reduced sequence read count error

1. A computer-implemented method for detecting presence or absence of a microdeletion or a microduplication, comprising:

(a) obtaining counts of sequence reads mapped to portions of a reference genome, which sequence reads are of circulating cell-free nucleic acid from a test sample from a pregnant female;

(b) reducing error in the counts of the sequence reads, wherein the error is reduced according to a process comprising:

(1) assigning a guanine and cytosine (GC) bias coefficient to the test sample based on a first fitted relation between (i) the counts of the sequence reads mapped to each of the portions and (ii) GC content for each of the portions, wherein the GC bias coefficient is a slope for a linear fitted relation for the first fitted relation or a curvature estimation for a non-linear fitted relation for the first fitted relation; and

(2) calculating a genomic section level for each of the portions for the test sample based on the counts of (a), the GC bias coefficient of (b)(1), and a second fitted relation, for each of the portions, between (i) the GC bias coefficient for each of multiple samples and (ii) the counts of the sequence reads mapped to each of the portions for the multiple samples, thereby providing calculated genomic section levels, whereby error in the counts of the sequence reads is reduced; and

(c) outputting a classification of the presence or absence of a microdeletion or a microduplication for the test sample according to the calculated genomic section levels, wherein a measure of deviation between (i) calculated genomic section levels for portions that include a microdeletion or a microduplication and (ii) expected genomic section levels for portions that include no microdeletion or microduplication is larger for counts with error reduction according to (b) than a measure of deviation between (i) and (ii) for counts with no error reduction according to (b).

2. The method of claim 1 , wherein the GC bias coefficient in (b)(1) is a slope for a linear fitted relation for the first fitted relation determined by linear regression.

3. The method of claim 1 , wherein the GC bias coefficient in (b)(1) is a curvature estimation determined by a non-linear fitted relation for the first fitted relation.

4. The method of claim 1 , wherein the second fitted relation of (b)(2) is linear.

5. The method of claim 4 , wherein a slope of the second fitted relation in (b)(2) is determined by linear regression.

6. The method of claim 5 , wherein the GC bias coefficient for each of the multiple samples in (b)(2)(i) is the slope of a third fitted linear relation, for each of the multiple samples, between (i) the counts of the sequence reads mapped to each of the portions and (ii) GC content for each of the portions.

7. The method of claim 6 , wherein a calculated genomic section level L is calculated for the test sample for each portion of the reference genome according to Equation B:

L =( M−GS )/ I   Equation B

wherein M is the counts of the sequence reads mapped to the portion for the test sample, G is the GC bias coefficient for the test sample, I is an intercept of the second fitted linear relation of (b)(2) for the portion, S is a slope of the second fitted linear relationship of (b)(2) for the portion.

8. The method of claim 1 , wherein (b)(1) or (b)(2), or (b)(1) and (b)(2), are implemented using at least one microprocessor.

9. The method of claim 1 , comprising after (b) generating at least one Z-score from the calculated genomic section levels.

10. The method of claim 1 , comprising filtering one or more portions and removing counts associated with the one or more portions for the classification of the presence or absence of the microdeletion or microduplication in part (c).

11. The method of claim 10 , comprising (i) normalizing the counts in (a) according to (b), thereby generating normalized counts, and removing normalized counts associated with one or more filtered portions, thereby yielding filtered normalized counts; or (ii) removing counts associated with one or more filtered portions prior to (b), and normalizing the counts in portions that were not removed according to (b), thereby yielding filtered normalized counts.

12. The method of claim 11 , comprising performing a secondary normalization of the filtered normalized counts.

13. The method of claim 12 , wherein the secondary normalization is a LOESS normalization.

14. The method of claim 10 , wherein the one or more filtered portions are selected according to one or more criteria chosen from measure of error or mappability, or measure of error and mappability.

15. The method of claim 14 , wherein the measure of error is an R factor.

16. The method of claim 15 , wherein portions of the reference genome having an R factor of about 7% or greater were selected as filtered portions.

17. The method of claim 16 , wherein portions of the reference genome having an R factor of about 7% to about 10% were selected as filtered portions.

18. The method of claim 10 , wherein the one or more filtered portions are selected according to one or more criteria chosen from portions having no guanosine and cytosine (GC) content, portions consistently receiving no counts, and repeat masking.

19. The method of claim 13 , comprising (i) identifying a copy number variation in the calculated genomic section levels, and (ii) adjusting the calculated genomic section levels associated with the copy number variation.

20. The method of claim 19 , comprising (1) identifying a first elevation in the calculated genomic section levels significantly different than a second elevation in the calculated genomic section levels, which first elevation is for a first set of genomic section levels, and which second elevation is for a second set of genomic section levels, (2) determining an expected elevation range for a homozygous and heterozygous copy number variation according to an uncertainty value for a segment of the genome, (3) adjusting the first elevation by a predetermined value when the first elevation is within one of the expected elevation ranges, thereby providing adjusted genomic section levels, and (4) identifying the presence or absence of the copy number variation according to the adjusted genomic section levels.

21. The method of claim 1 , comprising sequencing the test sample by a genome-wide massively parallel sequencing process, thereby generating the sequence reads.

22. The method of claim 21 , wherein the sequencing is at about 1-fold coverage or less.

23. The method of claim 21 , wherein the sequencing is at about 1-fold coverage or greater.

24. The method of claim 21 , comprising mapping the sequence reads to the portions of the reference genome, and counting the mapped sequence reads, thereby generating the counts of the sequence reads mapped to the portions of the reference genome.

25. The method of claim 1 , wherein the nucleic acid is from blood plasma or blood serum.