Integrated machine-learning framework to estimate homologous recombination deficiency
Methods, systems, and software are provided for determining a homologous recombination pathway status of a cancer in a test subject, e.g., to improve cancer treatment predictions and outcomes. In some embodiments, classifiers using one or more of (i) a heterozygosity status for DNA damage repair genes in a cancerous tissue, (ii) a measure of the loss of heterozygosity across the genome of the cancerous tissue, (iii) a measure of variant alleles detected in a second plurality of DNA damage repair genes in the genome of the cancerous tissue, (iv) a measure of variant alleles detected in the second plurality of DNA damage repair genes in the genome of a non-cancerous tissue, and (v) tumor sample purity are provided.
1 . A method of determining a homologous recombination deficiency (HRD) status of a test subject afflicted with a cancer, the method comprising:
at a computer system having one or more processors, and memory storing one or more programs for execution by the one or more processors:
(A) generating, based on a first plurality of sequence reads of a first DNA sample from a cancerous tissue of the subject and a second plurality of sequence reads of a second DNA sample from a non-cancerous tissue of the subject, a set of genomic data for the test subject comprising:
(i) a heterozygosity status for a first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the heterozygosity status derives from a count of a number of unique frameshift mutations or a number of unique truncating mutations detected in the first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject,
(ii) a measure of the loss of heterozygosity, on a proportionate scale, across the genome of the cancerous tissue of the subject,
(iii) a first measure of variant alleles detected in a second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the first measure of variant alleles derives from a first count of a number of unique mutations associated with a loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, and
(iv) a second measure of variant alleles detected in the second plurality of DNA damage repair genes in the genome of the non-cancerous tissue of the subject, wherein the second measure of variant alleles derives from a second count of a number of unique mutations associated with loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject;
(B) determining a measure of tumor purity in the test subject using the first plurality of sequence reads;
(C) using the tumor purity in the test subject to verify that the tumor purity in the test subject satisfies a tumor purity threshold;
(D) applying a scaling construct to the set of genomic data thereby obtaining a scaled set of genomic data, wherein the scaling construct is derived from normalization of the set of genomic data from each subject in a training cohort of subjects, wherein the training cohort of subjects includes a first subset of subjects having HRD and a second subject of subjects free of HRD; and
(E) inputting, after the using (C), the scaled set of genomic data into a classifier thereby determining a probability that the test subject has HRD.
2 . The method of claim 1 , wherein the first DNA sample is from a solid tumor biopsy of the cancerous tissue of the subject.
3 . The method of claim 1 , wherein the second DNA sample is from a buffy coat preparation of a blood sample from the subject.
4 . The method of claim 1 , the method further comprising performing targeted sequencing using a plurality of nucleic acid probes to enrich nucleic acids from the cancerous tissue of the subject for a panel of genomic regions to generate the first plurality of sequence reads.
5 . The method of claim 1 , the method further comprising performing whole genome sequencing of nucleic acids from the cancerous tissue of the subject to generate the first plurality of sequence reads.
6 . The method of claim 1 , the method further comprising performing targeted sequencing using a plurality of nucleic acid probes to enrich nucleic acids from the non-cancerous tissue of the subject for a panel of genomic regions to generate the second plurality of sequence reads.
7 . The method of claim 1 , the method further comprising performing whole genome sequencing of nucleic acids from the non-cancerous tissue of the subject to generate the second plurality of sequence reads.
8 . The method of claim 1 , the method further comprising determining the measure of the loss of heterozygosity, on the proportionate scale, across the genome of the cancerous tissue of the subject is determined by:
determining a loss of genomic heterozygosity in the first plurality of sequence reads, and
normalizing the determined loss of heterozygosity by an estimate of the tumor purity for the first plurality of sequence reads, wherein the estimate of the tumor purity is based on the first plurality of sequence reads and the second plurality of sequence reads.
9 . The method of claim 1 , wherein the first plurality of DNA damage repair genes comprises BRCA1 and BRCA2.
10 . The method of claim 1 , wherein the second plurality of DNA damage repair genes comprises BRCA1 and BRCA2.
11 . The method of claim 10 , wherein the unique mutations associated with loss of homologous recombination in BRCA1 and BRCA2 include at least 50 of the mutations selected from the group consisting of:
13_32890558_G_A, 13_32890593_TA_T, 13_32893214_A_C, 13_32893239_G_A, 13_32893373_C_A, 13_32893435_G_T, 13_32900288_G_T, 13_32900751_G_A, 13_32903604_CTG_C, 13_32905165_AT_A, 13_32906571_ATCTACAAAAAG_A, 13_32906909_G_T, 13_32907331_AGCTTT_A, 13_32907365_AAAAAG_A, 13_32907408_CATCTT_C, 13_32907420_G_GA, 13_32907420_GA_G, 13_32910644_G_T, 13_32910797_C_CT, 13_32911001_GA_G, 13_32911297_TAAAC_T, 13_32911321_T_TA, 13_32911356_AC_A, 13_32911380_T_TA, 13_32911470_G_A, 13_32911757_C_T, 13_32911775_C_T, 13_32911877_C_T, 13_32911968_GC_G, 13_32912089_CTG_C, 13_32912171_CTG_C, 13_32912351_AATAAT_A, 13_32912398_TG_T, 13_32912503_TG_T, 13_32912539_TCATA_T, 13_32912587_T_A, 13_32912701_TTCAAA T, 13_32912703_C_A, 13_32912735_G_T, 13_32912770_A_AT, 13_32912967_AAG_A, 13_32913118_GA_G, 13_32913139_AG_A, 13_32913143_C_T, 13_32913181_G_A, 13_32913457_C_G, 13_32913558_C_CA, 13_32913648_A_AT, 13_32913668_G_GA, 13_32913708_ATTTAAGT_A, 13_32913836_CA_C, 13_32914014_CA_C, 13_32914102_CAGTAA_C, 13_32914191_C_G, 13_32914209_ACT_A, 13_32914226_G_T, 13_32914247_A_T, 13_32914250_GT_G, 13_32914349_G_T, 13_32914437_GT_G, 13_32914502_G_T, 13_32914715_A_T, 13_32914757_G_T, 13_32914766_CTT_C, 13_32914851_C_A, 13_32915135_TACTC_T, 13_32915292_C_G, 13_32920978_C_T, 13_32929238_TCA_T, 13_32929275_G_T, 13_32929426_G_A, 13_32931878_G_A, 13_32936711_G_A, 13_32936732_G_C, 13_32936732_G_T, 13_32936828_C_A, 13_32936830_G_A, 13_32937354_T_TA, 13_32937354_TA_T, 13_32937479_CA_C, 13_32937506_G_C, 13_32944693_A_G, 13_32950809_AAC_A, 13_32950889_T_G, 13_32950928_G_A, 13_32950932_A_G, 13_32953453_G_A, 13_32953526_C_T, 13_32953556_G_T, 13_32953640_G_GA, 13_32953886_G_A, 13_32953974_C_G, 13_32954022_C_CA, 13_32954022_CA_C, 13_32954050_G_A, 13_32954147_TC T, 13_32954180_C_T, 13_32954222_C_T, 13_32954272_G_GA, 13_32954272_GA_G, 13_32968850_C_A, 13_32968863_C_G, 17_41197784_G_A, 17_41199658_A_T, 17_41203122_G_GC, 17_41209079_T_TG, 17_41209154_T_C, 17_41215362_TTTTC_T, 17_41215948_G_A, 17_41223097_G_A, 17_41223176_TG_T, 17_41223242_G_C, 17_41226411_G_A, 17_41226447_CTT_C, 17_41234421_CT_C, 17_41234451_G_A, 17_41243479_CTTGA_C, 17_41243533_C_A, 17_41243704_C_A, 17_41243800_C_A, 17_41243843_GTTTAC_G, 17_41244281_CA_C, 17_41244614_A_C, 17_41244865_GTT_G, 17_41244913_C_A, 17_41245161_G_GT, 17_41245203_CT_C, 17_41245330_CTT_C, 17_41245513_T_A, 17_41245586_CT_C, 17_41245603_C_A, 17_41245834_C_A, 17_41245888_C_A, 17_41246186_ACT_A, 17_41246197_AT_A, 17_41246278_CAG_C, 17_41246531_CT_C, 17_41246539_C_A, 17_41246723_G_GCCACATGGCT, 17_41247864_C_CT, 17_41247940_C_A, 17_41256203_TG_T, 17_41258504_A_C, 17_41267762_A_G, 17_41276044_ACT_A, and 17_41276048_TAA_T.
12 . The method of claim 10 , wherein the unique mutations associated with loss of homologous recombination in BRCA1 and BRCA2 comprises the mutations selected from the group consisting of:
13_32890558_G_A, 13_32890593_TA_T, 13_32893214_A_C, 13_32893239_G_A, 13_32893373_C_A, 13_32893435_G_T, 13_32900288_G_T, 13_32900751_G_A, 13_32903604_CTG_C, 13_32905165_AT_A, 13_32906571 ATCTACAAAAAG_A, 13_32906909_G_T, 13_32907331_AGCTTT_A, 13_32907365_AAAAAG_A, 13_32907408_CATCTT_C, 13_32907420_G_GA, 13_32907420_GA_G, 13_32910644_G_T, 13_32910797_C_CT, 13_32911001_GA_G, 13_32911297_TAAAC_T, 13_32911321_T_TA, 13_32911356_AC_A, 13_32911380_T_TA, 13_32911470_G_A, 13_32911757_C_T, 13_32911775_C_T, 13_32911877_C_T, 13_32911968_GC_G, 13_32912089_CTG_C, 13_32912171_CTG_C, 13_32912351_AATAAT_A, 13_32912398_TG_T, 13_32912503_TG_T, 13_32912539_TCATA_T, 13_32912587_T_A, 13_32912701_TTCAAA T, 13_32912703_C_A, 13_32912735_G_T, 13_32912770_A_AT, 13_32912967_AAG_A, 13_32913118_GA_G, 13_32913139_AG_A, 13_32913143_C_T, 13_32913181_G_A, 13_32913457_C_G, 13_32913558_C_CA, 13_32913648_A_AT, 13_32913668_G_GA, 13_32913708_ATTTAAGT_A, 13_32913836_CA_C, 13_32914014_CA_C, 13_32914102_CAGTAA_C, 13_32914191_C_G, 13_32914209_ACT_A, 13_32914226_G_T, 13_32914247_A_T, 13_32914250_GT_G, 13_32914349_G_T, 13_32914437_GT_G, 13_32914502_G_T, 13_32914715_A_T, 13_32914757_G_T, 13_32914766_CTT_C, 13_32914851_C_A, 13_32915135_TACTC_T, 13_32915292_C_G, 13_32920978_C_T, 13_32929238_TCA_T, 13_32929275_G_T, 13_32929426_G_A, 13_32931878_G_A, 13_32936711_G_A, 13_32936732_G_C, 13_32936732_G_T, 13_32936828_C_A, 13_32936830_G_A, 13_32937354_T_TA, 13_32937354_TA_T, 13_32937479_CA_C, 13_32937506_G_C, 13_32944693_A_G, 13_32950809_AAC_A, 13_32950889_T_G, 13_32950928_G_A, 13_32950932_A_G, 13_32953453_G_A, 13_32953526_C_T, 13_32953556_G_T, 13_32953640_G_GA, 13_32953886_G_A, 13_32953974_C_G, 13_32954022_C_CA, 13_32954022_CA_C, 13_32954050_G_A, 13_32954147_TC T, 13_32954180_C_T, 13_32954222_C_T, 13_32954272_G_GA, 13_32954272_GA_G, 13_32968850_C_A, 13_32968863_C_G, 17_41197784_G_A, 17_41199658_A_T, 17_41203122_G_GC, 17_41209079_T_TG, 17_41209154_T_C, 17_41215362_TTTTC_T, 17_41215948_G_A, 17_41223097_G_A, 17_41223176_TG_T, 17_41223242_G_C, 17_41226411_G_A, 17_41226447_CTT_C, 17_41234421_CT_C, 17_41234451_G_A, 17_41243479_CTTGA_C, 17 41243533_C_A, 17_41243704_C_A, 17_41243800_C_A, 17_41243843_GTTTAC_G, 17_41244281_CA_C, 17_41244614_A_C, 17_41244865_GTT_G, 17_41244913_C_A, 17_41245161_G_GT, 17_41245203_CT_C, 17_41245330_CTT_C, 17_41245513_T_A, 17_41245586_CT_C, 17_41245603_C_A, 17_41245834_C_A, 17_41245888_C_A, 17_41246186_ACT_A, 17_41246197_AT_A, 17_41246278_CAG_C, 17_41246531_CT_C, 17_41246539_C_A, 17_41246723_G_GCCACATGGCT, 17_41247864_C_CT, 17_41247940_C_A, 17_41256203_TG_T, 17_41258504_A_C, 17_41267762_A_G, 17_41276044_ACT_A, and 17_41276048_TAA_T.
13 . The method of claim 1 , wherein the method further comprises:
determining that the test subject has HRD; and
treating the cancer by administering a poly ADP ribose polymerase (PARP) inhibitor to the test subject.
14 . The method of claim 13 , wherein the PARP inhibitor is selected from the group consisting of olaparib, veliparib, rucaparib, niraparib, and talazoparib.
15 . The method of claim 1 , wherein the cancer is breast cancer.
16 . The method of claim 1 , wherein the cancer is ovarian cancer.
17 . The method of claim 1 , wherein the cancer is colorectal cancer.
18 . The method of claim 1 , wherein the classifier is a neural network, a support vector machine, a Naive Bayes classifier, a nearest neighbor classifier, a boosted trees classifier, a random forest classifier, a convolutional neural network classifier, a decision tree classifier, a regression classifier, or a clustering classifier.
19 . The method of claim 1 , wherein the classifier is a random forest classifier.
20 . The method of claim 1 , the method further comprising exome sequencing a plurality of cDNA molecules generated from the cancerous tissue of the subject to generate the first plurality of sequence reads.
21 . The method of claim 1 , the method further comprising exome sequencing a plurality of cDNA molecules generated from the non-cancerous tissue of the subject to generate the second plurality of sequence reads.
22 . The method of claim 1 , wherein the classifier determines that the probability that test subject has HRD is greater than fifty percent, and the method further comprises positively calling the test subject as HRD.
23 . The method of claim 1 , wherein the classifier determines that the probability that test subject has HRD is less than fifty percent, and the method further comprises calling the test subject as free of HRD.
24 . A computer system comprising:
one or more processors; and
a non-transitory computer-readable medium including computer-executable instructions that, when executed by the one or more processors, cause the processors to perform a method of determining a homologous recombination deficiency (HRD) status of a test subject afflicted with a cancer, the method comprising:
(A) generating, based on a first plurality of sequence reads of a first DNA sample from a cancerous tissue of the subject and a second plurality of sequence reads of a second DNA sample from a non-cancerous tissue of the subject, a set of genomic data for the test subject comprising:
(i) a heterozygosity status for a first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the heterozygosity status derives from a count of a number of unique frameshift mutations or a number of unique truncating mutations detected in the first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject,
(ii) a measure of the loss of heterozygosity, on a proportionate scale, across the genome of the cancerous tissue of the subject,
(iii) a first measure of variant alleles detected in a second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the first measure of variant alleles derives from a first count of a number of unique mutations associated with a loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, and
(iv) a second measure of variant alleles detected in the second plurality of DNA damage repair genes in the genome of the non-cancerous tissue of the subject, wherein the second measure of variant alleles derives from a second count of a number of unique mutations associated with loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject;
(B) determining a measure of tumor purity in the test subject using the first plurality of sequence reads;
(C) using the tumor purity in the test subject to verify that the tumor purity in the test subject satisfies a tumor purity threshold;
(D) applying a scaling construct to the set of genomic data thereby obtaining a scaled set of genomic data, wherein the scaling construct is derived from normalization of the set of genomic data from each subject in a training cohort of subjects, wherein the training cohort of subjects includes a first subset of subjects having HRD and a second subject of subjects free of HRD; and
(E) inputting, after the using (C), the scaled set of genomic data into a classifier thereby determining a probability that the test subject has HRD.
25 . A non-transitory computer-readable storage medium having stored thereon program code instructions that, when executed by a processor, cause the processor to perform a method of determining a homologous recombination deficiency (HRD) status of a test subject afflicted with a cancer, the method comprising:
(A) generating, based on a first plurality of sequence reads of a first DNA sample from a cancerous tissue of the subject and a second plurality of sequence reads of a second DNA sample from a non-cancerous tissue of the subject, a set of genomic data for the test subject comprising:
(i) a heterozygosity status for a first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the heterozygosity status derives from a count of a number of unique frameshift mutations or a number of unique truncating mutations detected in the first plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject,
(ii) a measure of the loss of heterozygosity, on a proportionate scale, across the genome of the cancerous tissue of the subject,
(iii) a first measure of variant alleles detected in a second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, wherein the first measure of variant alleles derives from a first count of a number of unique mutations associated with a loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject, and
(iv) a second measure of variant alleles detected in the second plurality of DNA damage repair genes in the genome of the non-cancerous tissue of the subject, wherein the second measure of variant alleles derives from a second count of a number of unique mutations associated with loss of homologous recombination detected in the second plurality of DNA damage repair genes in the genome of the cancerous tissue of the subject:
(B) determining a measure of tumor purity in the test subject using the first plurality of sequence reads;
(C) using the tumor purity in the test subject to verify that the tumor purity in the test subject satisfies a tumor purity threshold;
(D) applying a scaling construct to the set of genomic data thereby obtaining a scaled set of genomic data, wherein the scaling construct is derived from normalization of the set of genomic data from each subject in a training cohort of subjects, wherein the training cohort of subjects includes a first subset of subjects having HRD and a second subject of subjects free of HRD; and
(E) inputting, after the using (C), the scaled set of genomic data into a classifier thereby determining a probability that the test subject has HRD.