Limit of detection based quality control metric
Provided are methods and systems for sample quality control in CNV detection using test samples comprising cell-free nucleic acid fragments originating from a mother and a fetus. The method involves determining an exclusion region defined by at least a fetal fraction limit of detection (LOD) curve. The fetal fraction LOD curve varies with coverage values and indicates minimum values of fetal fractions needed to achieve a detection criterion given different coverages.
1. A method, implemented using a nucleic acid sequencer and a computer system comprising one or more processors and memory, for processing a test sample comprising cell-free nucleic acid fragments originating from a mother and a fetus, the method comprising:
(A) sequencing, at a first depth of sequencing, the test sample comprising the cell-free nucleic acid fragments originating from a mother and a fetus;
(B) computationally analyzing sequence information from the test sample, the analyzing comprising:
(a) determining an observed value of a fetal fraction of the test sample, wherein the fetal fraction of the test sample indicates a relative amount of fetal origin cell-free nucleic acid fragments in the test sample;
(b) receiving, by the computer system, sequence reads obtained by sequencing the cell-free nucleic acid fragments in the test sample;
(c) aligning, by the computer system, the sequence reads of the cell-free nucleic acid fragments to a reference genome comprising a sequence of interest, thereby providing sequence tags;
(d) determining, by the computer system, a coverage of the sequence tags for at least a portion of the reference genome; and
(e) determining that the test sample is within an exclusion region based on the coverage of sequences tags determined in (d) and the observed value of the fetal fraction determined in (a), wherein the exclusion region is defined by at least a fetal fraction limit of detection (LOD) curve, wherein the fetal fraction LOD curve varies with coverage values and indicates, for a given coverage value of different coverage values, a minimum observed value of the fetal fraction to achieve a given level of confidence that a corresponding ground truth fetal fraction is greater than a minimum value of fetal fractions needed to achieve a detection criterion; and
(C) re-sequencing, at a second depth of sequencing that is greater than the first depth of sequencing, the test sample to obtain re-sequenced sequence reads;
(D) repeating (B) (a)-(d) for the re-sequenced sequence reads; and
(E) determining that the re-sequenced test sample is outside the exclusion region;
further comprising: obtaining the LOD curve by:
(i) selecting from a first plurality of training samples, for each observed fetal fraction of a plurality of observed fetal fractions and for each coverage level of a plurality of coverage levels, one or more selected training samples whose true fetal fractions have the given level of confidence, wherein each training sample comprises cell-free nucleic acid fragments originating from a mother and a fetus,
(ii) obtaining, using the one or more selected training samples for each observed fetal fraction and each coverage level, true fetal fraction versus observed fetal fraction curves of the plurality of coverage levels,
(iii) obtaining, using a second plurality of training samples, true fetal fractions having at least the minimum value of fetal fraction needed to achieve the detection criterion for the plurality of coverage levels,
(iv) obtaining, using the true fetal fraction versus observed fetal fraction curves and the true fetal fractions having at least the minimum value of fetal fraction needed to achieve the detection criterion, a plurality of minimal observed fetal fractions for the plurality of coverage levels required to have the given level of confidence that true fetal fractions have at least the minimum value of fetal fraction needed to achieve the detection criterion, and
(v) obtaining the fetal fraction LOD curve using the plurality of minimal observed fetal fractions for the plurality of coverage levels.
2. The method of claim 1 , further comprising, after (B) and prior to (C), determining that the test sample is negative for a copy number variation (CNV) of the sequence of interest.
3. The method of claim 1 , further comprising:
repeating (a)-(d) using the re-sequenced sequence reads;
determining that the test sample is outside the exclusion region; and
(F) in response to determining that the re-sequenced test sample is outside the exclusion region, calling the test sample as either having a copy number variation (CNV) of the sequence of interest or not having the CNV of the sequence of interest.
4. The method of claim 1 , wherein the fetal fraction LOD curve is obtained based on LODs of affected training samples that are affected by a copy number variation (CNV).
5. The method of claim 1 , wherein the detection criterion is X % confident that, for the minimum observed value of the fetal fraction, the corresponding ground truth fetal fraction is larger than a specified LOD needed to detect the affected samples with Y % confidence of detection.
6. The method of claim 5 , wherein the X % is about 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 99%, or 99.5%.
7. The method of claim 5 , wherein the Y % is about 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, or 99% confidence of detection.
8. The method of claim 5 , wherein the X % is 50% and the Y % is 95%.
9. The method of claim 1 , wherein the minimum observed value of the fetal fraction is determined as a smallest observed fetal fraction at which Y % of the affected samples can be detected.
10. The method of claim 1 , wherein the detection criterion for an observed fetal fraction at an observed coverage is obtained using a distribution of ground truth fetal fractions of the observed fetal fraction at the observed coverage.
11. The method of claim 1 , wherein the exclusion region is under the fetal fraction LOD curve.
12. The method of claim 1 , wherein the exclusion region is under both the fetal fraction LOD curve and a coverage threshold.
13. The method of claim 1 , wherein the observed value of the fetal fraction of the test sample is determined based on sizes of the cell-free nucleic acid fragments.
14. The method of claim 13 , wherein the observed value of the fetal fraction of the test sample is determined by:
obtaining a frequency distribution of the sizes of the cell-free nucleic acid fragments; and
applying the frequency distribution to a model relating fetal fraction to frequency of fragment size to obtain the observed value of the fetal fraction.
15. The method of claim 1 , wherein the observed value of the fetal fraction of the test sample is determined based on coverage information for bins of the reference genome.
16. The method of claim 15 , wherein the observed value of the fetal fraction is calculated by: applying coverage values of a plurality of bins of the reference genome to a model relating fetal fraction to coverage of a bin to obtain the observed value of the fetal fraction.
17. The method of claim 1 , wherein the observed value of the fetal fraction of the test sample is determined based on coverage information for bins of a sex chromosome.
18. A computer system for evaluation of copy number of a nucleic acid sequence of interest in a test sample comprising cell-free nucleic acid fragments originating from a mother and a fetus, the system comprising a sequencer and a computer system comprising a processor and one or more computer-readable storage media having stored thereon instructions for execution on said processor, wherein the system is configured to cause:
(A) the sequencer to sequence, at a first depth of sequencing, the test sample;
(B) the computer system to:
(a) determine an observed value of a fetal fraction of the test sample, wherein the fetal fraction of the test sample indicates a relative amount of fetal origin cell-free nucleic acid fragments in the test sample;
(b) receive sequence reads obtained by sequencing the cell-free nucleic acid fragments in the test sample;
(c) align the sequence reads of the cell-free nucleic acid fragments to a reference genome comprising a sequence of interest, thereby providing sequence tags;
(d) determine a coverage of the sequence tags for at least a portion of the reference genome; and
(e) determine that the test sample is within an exclusion region based on the coverage of sequences tags determined in (d) and the observed value of the fetal fraction determined in (a), wherein the exclusion region is defined by at least a fetal fraction limit of detection (LOD) curve, wherein the fetal fraction LOD curve varies with coverage values and indicates, for a given coverage value of different coverage values, a minimum observed value of the fetal fraction to achieve a given level of confidence that a corresponding ground truth fetal fraction is greater than a minimum value of fetal fractions needed to achieve a detection criterion; and
(C) the sequencer to re-sequence, at a second depth of sequencing that is greater than the first depth of sequencing, the test sample; and
(D) the computer system to repeat (a)-(d) for the re-sequenced sequence reads;
further comprising: obtaining the LOD curve by:
(i) selecting from a first plurality of training samples, for each observed fetal fraction of a plurality of observed fetal fractions and for each coverage level of a plurality of coverage levels, one or more selected training samples whose true fetal fractions have the given level of confidence, wherein each training sample comprises cell-free nucleic acid fragments originating from a mother and a fetus,
(ii) obtaining, using the one or more selected training samples for each observed fetal fraction and each coverage level, true fetal fraction versus observed fetal fraction curves of the plurality of coverage levels,
(iii) obtaining, using a second plurality of training samples, true fetal fractions having at least the minimum value of fetal fraction needed to achieve the detection criterion for the plurality of coverage levels,
(iv) obtaining, using the true fetal fraction versus observed fetal fraction curves and the true fetal fractions having at least the minimum value of fetal fraction needed to achieve the detection criterion, a plurality of minimal observed fetal fractions for the plurality of coverage levels required to have the given level of confidence that true fetal fractions have at least the minimum value of fetal fraction needed to achieve the detection criterion, and
(v) obtaining the fetal fraction LOD curve using the plurality of minimal observed fetal fractions for the plurality of coverage levels.
19. The system of claim 18 , wherein the system is further configured to cause:
(E) the computer system to, in response to determining that the re-sequenced test sample is outside the exclusion region, make a call of a copy number variation (CNV) of the sequence of interest.