Statistical modeling for sensitive detection of low-frequency single nucleotide variants.

paper Primary Public

Authors: Hao, Yangyang; Zhang, Pengyue; Xuei, Xiaoling; Nakshatri, Harikrishna; Edenberg, Howard J; Li, Lang; Liu, Yunlong
Year: 2016
Journal: BMC genomics
PMID: 27556804
DOI: 10.1186/s12864-016-2905-x
PMCID: PMC5001245

BACKGROUND: Sensitive detection of low-frequency single nucleotide variants carries great significance in many applications. In cancer genetics research, tumor biopsies are a mixture of normal and tumor cells from various subpopulations due to tumor heterogeneity. Thus the frequencies of somatic variants from a subpopulation tend to be low. Liquid biopsies, which monitor circulating tumor DNA in blood to detect metastatic potential, also face the challenge of detecting low-frequency variants due to the small percentage of the circulating tumor DNA in blood. Moreover, in population genetics research, although pooled sequencing of a large number of individuals is cost-effective, pooling dilutes the signals of variants from any individual. Detection of low frequency variants is difficult and can be cofounded by sequencing artifacts. Existing methods are limited in sensitivity and mainly focus on frequencies around 2 % to 5 %; most fail to consider differential sequencing artifacts. RESULTS: We aimed to push down the frequency detection limit close to the position specific sequencing error rates by modeling the observed erroneous read counts with respect to genomic sequence contexts. 4 distributions suitable for count data modeling (using generalized linear models) were extensively characterized in terms of their goodness-of-fit as well as the performances on real sequencing data benchmarks, which were specifically designed for testing detection of low-frequency variants; two sequencing technologies with significantly different chemistry mechanisms were used to explore systematic errors. We found the zero-inflated negative binomial distribution generalized linear mode is superior to the other models tested, and the advantage is most evident at 0.5 % to 1 % range. This method is also generalizable to different sequencing technologies. Under standard sequencing protocols and depth given in the testing benchmarks, 95.3 % recall and 79.9 % precision for Ion Proton data, 95.6 % recall and 97.0 % precision for Illumina MiSeq data were achieved for SNVs with frequency > = 1 %, while the detection limit is around 0.5 %. CONCLUSIONS: Our method enables sensitive detection of low-frequency single nucleotide variants across different sequencing platforms and will facilitate research and clinical applications such as pooled sequencing, cancer early detection, prognostic assessment, metastatic monitoring, and relapses or acquired resistance identification.

#	Section	Preview
0	Background	In 2005, the first next-generation sequencing (NGS) technology was released by 454 Life Sciences…
1	Background	In cancer genetics research, low frequency tumor somatic SNV identification is crucial due to the…
2	Background	The difficulty for low-frequency SNV identification using NGS technologies is due to the relatively…
3	Background	for substitution types while the latter only utilized information in one location thus the…
4	Background	By analyzing previous efforts, our group proposed a framework (Yangyang Hao XX, Li L, Nakshatri H,…
5	Results	In the Result section, we first briefly summarized the benchmark datasets used for PSEM. Then we…
6	Results — Benchmarks overview and comparison	Two sets of designed benchmarks targeting low-frequency SNVs from both Ion Proton and Illumina MiSeq…
7	Results — Benchmarks overview and comparison	Comparing the two testing benchmarks, Ion Proton contains a total of 1557 somatic SNVs while…
8	Results — Benchmarks overview and comparison	Except for allele frequency composition, sequencing depth is also a crucial factor affecting the…
9	Results — Candidate distributions selection	To model error rate based on count data, 3 most common distribution choices are binomial, Poisson…
10	Results — Comparing the goodness-of-fit of different distributions	9 genomic sequence context covariates, totaling 24 degrees of freedom, were included in the GLM…
11	Results — Comparing the goodness-of-fit of different distributions	To compare the goodness-of-fit of models based on different distributions, we used Vuong’s…
12	Results — Comparing the goodness-of-fit of different distributions	considering both dispersion and zero-inflation is further emphasized by the comparisons of ZIP with…
13	Results — Performance evaluation on Ion Proton testing benchmark	We first evaluated the overall precision and recall values of all models on the test benchmark. From…
14	Results — Performance evaluation on Ion Proton testing benchmark	frequency, whereas the majority of previous studies focused on SNVs of > = 5 % allele frequency; 2.…
15	Results — Performance evaluation on Ion Proton testing benchmark	Next, for all distributions, we explored the performance profiles on different allele frequencies.…
16	Results — Performance evaluation on Ion Proton testing benchmark	which pinpoints the necessity of modeling zero-inflation to derive more accurate error rates…
17	Results — Application of ZINB PSEM on Illumina MiSeq data	To evaluate the generalizability and adaptiveness of the GLM based PSEM, the same modeling…
18	Results — Application of ZINB PSEM on Illumina MiSeq data	Despite similar statistical modeling schema can be readily generalized to Illumina MiSeq data set,…
19	Results — Application of ZINB PSEM on Illumina MiSeq data	To evaluate whether the differences in GLM coefficients affect the performance profiles on various…

Name	Type
0.5% frequency SNV local	variant
1000 Genomes Project	cohort
1% frequency SNV local	variant
acquired resistance local	phenotype
benchmark datasets local	cohort
Binomial distribution local	drug
blood DNA local	drug
CAL_A local	cohort
CAL_A dataset local	cohort
CAL_B local	cohort
CAL_C local	cohort
CAL_D local	cohort
cancer	phenotype
candidate SNV local	variant
candidate SNVs local	variant
circulating tumor DNA local	drug
DNA	drug
DNA-Seq local	drug
duplex sequencing	drug
early detection local	phenotype
EdgeR local	drug
Fisher’s exact test local	drug
homopolymer related errors local	phenotype
Illumina MiSeq local	cohort
Illumina MiSeq local	drug
Illumina MiSeq benchmark local	cohort
Illumina MiSeq dataset local	cohort
invariant loci local	variant
Ion Proton local	cohort
Ion Proton local	drug
Ion Proton dataset local	cohort
Ion Proton sequencing dataset local	cohort
Ion Proton test benchmark local	cohort
Ion Proton test benchmark dataset local	cohort
Ion Proton training benchmark local	cohort
Ion Proton training dataset local	cohort
low-frequency SNVs local	variant
low frequency tumor somatic SNV local	variant
metastasis	phenotype
metastatic monitoring local	phenotype
MiSeq local	cohort
MiSeq training data local	cohort
NA11993 local	cohort
NA12878	cohort
NB local	drug
NB distribution local	drug
NB GLM local	drug
Negative binomial distribution local	drug
Poisson local	drug
Poisson distribution	drug
Poisson GLM local	drug
prognostic assessment local	phenotype
PSEM local	drug
PSEM approach local	drug
PSEM framework local	drug
PSEM model local	drug
relapse	phenotype
RNA-seq	drug
sequencing depth local	drug
sequencing depth evenness local	phenotype
single nucleotide variant	variant
SNV	variant
SNVs	variant
somatic SNV local	variant
SRP009487.1 local	cohort
testing benchmark local	cohort
training benchmark local	cohort
UDT-Seq local	drug
ultra-deep target enrichment assay local	drug
VarScan2 local	drug
Vuong’s non-nested test local	drug
Vuong’s test local	drug
Zero-inflated negative binomial local	drug
Zero-inflated Poisson local	drug
ZINB local	drug
ZINB GLM local	drug
ZIP local	drug

Citation	PMID	DOI	Status
Bragg, LM et al., PLoS Comput Biol, 2013, Shining a light on dark sequencing: characterising errors in Ion Torrent PGM data	23592973	10.1371/journal.pcbi.1003031	Cited
Cameron, AC et al., J Econometrics, 1990, Regression-Based Tests for Overdispersion in the Poisson Model	—	10.1016/0304-4076(90)90014-K	Cited
Carter, SL et al., Nat Biotechnol, 2012, Absolute quantification of somatic DNA alterations in human cancer	22544022	10.1038/nbt.2203	Cited
Cibulskis, K et al., Nat Biotechnol, 2013, Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples	23396013	10.1038/nbt.2514	Cited
Crowley, E et al., Nat Rev Clin Oncol, 2013, Liquid biopsy: monitoring cancer-genetics in the blood	23836314	10.1038/nrclinonc.2013.110	Cited
Diehl, F et al., Proc Natl Acad Sci U S A, 2005, Detection and quantification of mutations in the plasma of patients with colorectal tumors	16258065	10.1073/pnas.0507904102	Cited
Friendly, M, Visualizing Categorical Data, 2000	—	—	—
Goya, R et al., Bioinformatics, 2010, SNVMix: predicting single nucleotide variants from next-generation sequencing of tumors	20130035	10.1093/bioinformatics/btq040	Cited
Greene WH. Accounting for Excess Zeros and Sample Selection in Poisson and Negative Binomial Regression Models. In: NYU Working Paper No. EC-94-10; 1994.	—	—	—
Harismendy, O et al., Genome Biol, 2011, Detection of low prevalence somatic mutations in solid tumors with ultra-deep targeted sequencing	22185227	10.1186/gb-2011-12-12-r124	Cited
Hoaglin, DC et al., Checking the Shape of Discrete Distributions, in Exploring Data Tables, Trends, and Shapes, 2011, Checking the Shape of Discrete Distributions	—	—	—
Hoaglin, DC, Am Stat, 1980, A poissonness plot	—	—	—
Kennedy, SR et al., Nat Protoc, 2014, Detecting ultralow-frequency mutations by Duplex Sequencing	25299156	10.1038/nprot.2014.170	Cited
Koboldt, DC et al., Genome Res, 2012, VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing	22300766	10.1101/gr.129684.111	Cited
Lambert, D, Technometrics, 1992, Zero-Inflated Poisson Regression, with an Application to Defects in Manufacturing	—	10.2307/1269547	Cited
Li, H, Bioinformatics, 2011, A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data	21903627	10.1093/bioinformatics/btr509	Cited
Margulies, M et al., Nature, 2005, Genome sequencing in microfabricated high-density picolitre reactors	16056220	10.1038/nature03959	Cited
McElroy, KE et al., BMC Genomics, 2012, GemSIM: general, error-model based simulator of next-generation sequencing data	22336055	10.1186/1471-2164-13-74	Cited
Meacham, CE et al., Nature, 2013, Tumour heterogeneity and cancer cell plasticity	24048065	10.1038/nature12624	Cited
Robinson, MD et al., Bioinformatics, 2010, edgeR: a Bioconductor package for differential expression analysis of digital gene expression data	19910308	10.1093/bioinformatics/btp616	Cited
Ross, MG et al., Genome Biol, 2013, Characterizing and measuring bias in sequence data	23718773	10.1186/gb-2013-14-5-r51	Cited
Saunders, CT et al., Bioinformatics, 2012, Strelka: accurate somatic small-variant calling from sequenced tumor-normal sample pairs	22581179	10.1093/bioinformatics/bts271	Cited
Schmitt, MW et al., Nat Methods, 2015, Sequencing small genomic targets with high efficiency and extreme accuracy	25849638	10.1038/nmeth.3351	Cited
van Dijk, EL et al., Trends in Genetics: TIG, 2014, Ten years of next-generation sequencing technology	25108476	10.1016/j.tig.2014.07.001	Cited
Vuong, QH, Econometrica, 1989, Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses	—	10.2307/1912557	Cited

Title	Authors	Journal	Year	Link
GeneBits: ultra-sensitive tumour-informed ctDNA monitoring of treatment response and relapse in cancer patients.	Broche J et al.	—	2025	→
A system for detecting high impact-low frequency mutations in primary tumors and metastases.	Anjanappa M et al.	—	2018	→
RareVar: A Framework for Detecting Low-Frequency Single-Nucleotide Variants.	Hao Y et al.	—	2017	→
Intelligent biology and medicine in 2015: advancing interdisciplinary education, collaboration, and data science.	Huang K et al.	—	2016	→

Statistical modeling for sensitive detection of low-frequency single nucleotide variants.

External