Method and system of cancer detection using CpG-SNP contamination markers
Methods and systems for detecting contaminated fragments in a biological sample for cancer classification are disclosed. The system identifies CpG-SNP contamination markers. The CpG-SNP contamination markers include at least one SNP that affects a CpG site. The CpG-SNP contamination markers may include additive CpG-SNP sites and/or subtractive CpG-SNP sites. Additive CpG-SNP sites include an SNP that creates a new CpG site. Subtractive CpG-SNP sites include an SNP that removes a preexisting CpG site. Hybrid sites may include additional sites. A multiple CpG-SNP contamination marker comprises two or more CpG-SNP sites. A CpG-SNP & indel contamination marker comprises at least one CpG-SNP site and an indel site. For a given sample, the system identifies contamination markers for which the sample is homozygous. The system determines fragments having a haplotype that is different from the homozygous haplotype of the sample to be contamination fragments.
1 . A method for predicting a presence of cancer in a test sample, the method comprising:
i) obtaining the test sample comprising a plurality of cell-free DNA (cfDNA) fragments, the test sample collected from an individual;
ii) sequencing the cfDNA fragments with a plurality of probes designed to target a plurality of genetic markers yielding a plurality of sequence reads, the plurality of genetic markers including a plurality of CpG single nucleotide polymorphism (CpG-SNP) contamination markers for identifying any foreign fragments in the test sample belonging not to the individual associated with the test sample, wherein at least one of the plurality of CpG-SNP contamination markers comprises:
(a) an additive CpG-SNP site, wherein the SNP at the first CpG-SNP contamination marker creates a new CpG site, and wherein the first CpG-SNP contamination marker is one of: a thymine-cytosine polymorphism in a thymine-guanine dinucleotide, an adenine-cytosine polymorphism in an adenine-guanine dinucleotide, a guanine-cytosine polymorphism in a guanine-guanine dinucleotide, a thymine-guanine polymorphism in a cytosine-thymine dinucleotide, an adenine-guanine polymorphism in a cytosine-adenine dinucleotide, and a cytosine-guanine polymorphism in a cytosine-cytosine dinucleotide; or
(b) a subtractive CpG-SNP site, wherein the SNP at the first CpG-SNP contamination marker removes a preexisting CpG site, and wherein the first CpG-SNP contamination marker is one of: a cytosine-thymine polymorphism in a cytosine-guanine dinucleotide; a cytosine-adenine polymorphism in a cytosine-guanine dinucleotide; a cytosine-guanine polymorphism in a cytosine-guanine dinucleotide; a guanine-thymine polymorphism in a cytosine-guanine dinucleotide; a guanine-adenine polymorphism in a cytosine-guanine dinucleotide; and a guanine-cytosine polymorphism in a cytosine-guanine dinucleotide;
iii) identifying, based on the sequence reads, one or more CpG-SNP contamination markers from the plurality of CpG-SNP contamination markers for which the test sample has a homozygous haplotype;
iv) for each of the identified one or more CpG-SNP contamination markers for which the test sample has a homozygous haplotype, determining whether the sequence reads for the cfDNA fragments have a different haplotype at the identified CpG-SNP contamination marker than the homozygous haplotype of the test sample, wherein cfDNA fragments having a different haplotype at the identified CpG-SNP contamination marker than the homozygous haplotype of the test sample are labeled as contamination cfDNA fragments originating from another source that is not the individual;
v) determining if the test sample is contaminated based on a number of contamination cfDNA fragments being below a threshold; and
vi) in response to determining that the test sample is contaminated, excluding the sample from further analysis;
vii) in response to determining that the test sample is not contaminated, applying a classification model to the sequence reads for the cfDNA fragments excluding any sequence reads associated with contamination cfDNA fragments and outputting a cancer prediction for the test sample based on the sequence reads for the cfDNA fragments.
2 . The method of claim 1 , wherein applying the classification model comprises: generating a test feature vector based on the sequence reads excluding any sequence reads associated with contamination cfDNA fragments; inputting the test feature vector into the classification model; and generating a cancer prediction for the test sample.
3 . The method of claim 1 , wherein a given CpG-SNP contamination marker has a population methylation frequency above a threshold frequency, and wherein the threshold frequency is selected from the range of 70%-100%.
4 . The method of claim 1 , wherein each CpG-SNP contamination marker has population haplotype frequencies within the range of 45%-55%.
5 . The method of claim 1 , wherein the haplotypes of each CpG-SNP contamination marker are in Hardy-Weinberg equilibrium.
6 . The method of claim 1 , wherein a given CpG-SNP contamination marker further comprises a second SNP affecting a second CpG site.
7 . The method of claim 6 , wherein the second CpG site and the first CpG site of the given CpG-SNP contamination marker are within a threshold distance, and wherein the threshold distance is selected from the range of 8 bp to 30 bp.
8 . The method of claim 6 , wherein: the first SNP removes the first CpG site and the second SNP removes the second CpG site, the first SNP removes the first CpG site and the second SNP creates the second CpG site, or the first SNP creates the first CpG site and the second SNP creates the second CpG site.
9 . The method of claim 6 , wherein the given CpG SNP marker has population haplotype frequencies within the range of 45%-55%.
10 . The method of claim 6 , wherein the haplotypes of the given CpG SNP marker are in Hardy-Weinberg equilibrium.
11 . The method of claim 6 , wherein the given SNP CpG marker further comprises an insertion or a deletion (indel) sequence within a threshold distance from the first CpG site and the second CpG site, and wherein the threshold distance is selected from 100 bp to 5 kbp.
12 . The method of claim 11 , wherein the insertion indel sequence is of a threshold length, and wherein the threshold length is selected from 5 bp to 30 bp.
13 . The method of claim 12 , wherein the insertion indel sequence affects a third CpG site.
14 . The method of claim 1 , wherein each contamination marker includes a probe designed to target each haplotype of the contamination marker.
15 . The method of claim 1 , wherein the cancer prediction is a binary prediction between cancer and non-cancer.
16 . The method of claim 1 , further comprising: filtering, after said obtaining, the cfDNA fragments of the test sample to generate a set of anomalous cfDNA fragments, wherein said filtering comprises removing cfDNA fragments having below a threshold p-value with respect to other fragments.
17 . A system for predicting a presence of cancer in a test sample comprising:
a sequencing device configured to: i) load a test sample comprising a plurality of cell-free DNA (cfDNA) fragments, the test sample collected from an individual; ii) sequence the cfDNA fragments with a plurality of probes designed to target a plurality of genetic markers yielding a plurality of sequence reads, the plurality of genetic markers including a plurality of CpG single nucleotide polymorphism (CpG-SNP) contamination markers for identifying any foreign fragments in the test sample belonging not to the individual associated with the test sample, wherein at least one of the plurality of CpG-SNP contamination markers comprises:
(a) an additive CpG-SNP site, wherein the SNP at the first CpG-SNP contamination marker creates a new CpG site, and wherein the first CpG-SNP contamination marker is one of: a thymine-cytosine polymorphism in a thymine-guanine dinucleotide, an adenine-cytosine polymorphism in an adenine-guanine dinucleotide, a guanine-cytosine polymorphism in a guanine-guanine dinucleotide, a thymine-guanine polymorphism in a cytosine-thymine dinucleotide, an adenine-guanine polymorphism in a cytosine-adenine dinucleotide, and a cytosine-guanine polymorphism in a cytosine-cytosine dinucleotide; or
(b) a subtractive CpG-SNP site, wherein the SNP at the first CpG-SNP contamination marker removes a preexisting CpG site, and wherein the first CpG-SNP contamination marker is one of: a cytosine-thymine polymorphism in a cytosine-guanine dinucleotide; a cytosine-adenine polymorphism in a cytosine-guanine dinucleotide; a cytosine-guanine polymorphism in a cytosine-guanine dinucleotide; a guanine-thymine polymorphism in a cytosine-guanine dinucleotide; a guanine-adenine polymorphism in a cytosine-guanine dinucleotide; and a guanine-cytosine polymorphism in a cytosine-guanine dinucleotide;
a computer processor and a non-transitory computer-readable storing instructions that, when executed by the computer processor, cause the computer processor to perform operations comprising:
iii) identifying, based on the sequence reads, one or more CpG-SNP contamination markers from the plurality of CpG-SNP contamination markers for which the test sample has a homozygous haplotype;
iv) for each of the identified one or more CpG-SNP contamination markers for which the test sample has a homozygous haplotype, determining whether the sequence reads for the cfDNA fragments have a different haplotype at the identified CpG-SNP contamination marker than the homozygous haplotype of the test sample, wherein cfDNA fragments having a different haplotype at the identified CpG-SNP contamination marker than the homozygous haplotype of the test sample are labeled as contamination cfDNA fragments originating from another source that is not the individual;
v) determining if the test sample is contaminated based on a number of contamination cfDNA fragments being below a threshold; and
vi) in response to determining that the test sample is contaminated, excluding the sample from further analysis;
vii) in response to determining that the test sample is not contaminated, applying a classification model to the sequence reads for the cfDNA fragments excluding any sequence reads associated with contamination cfDNA fragments and outputting a cancer prediction for the test sample based on the sequence reads for the cfDNA fragments.
18 . The method of claim 1 , wherein sequencing the cfDNA fragments yielding the plurality of sequence reads comprises: treating the cfDNA fragments to convert any unmethylated cytosines to uracils; and sequencing the treated cfDNA fragments to identify any uracils in the treated cfDNA fragments indicating unmethylation at a CpG site and to identify any unconverted cytosines in the treated cfDNA fragments indicating methylation at a CpG site.