M3-S: a genotype calling method incorporating information from samples with known genotypes

BMC Bioinformatics

Methodology Article

Hongyu Zhao¹  Gengxin Li² 

[1] Department of Biostatistics, Yale School of Public Health, 60 College Street, 06520, New Haven, USA;Department of Mathematics and Statistics, Wright State University, 3640 Colonel Glenn Hwy, 45435, Dayton, USA;

关键词: Gaussian mixture model (GMM);    Clustering;    Genotype;    Genotyping array;    HapMap;    Single nucleotide polymorphisms (SNPs);    Rare SNP;   

DOI  :  10.1186/s12859-015-0824-5

 received in 2015-04-13, accepted in 2015-11-02,  发布年份 2015

来源: Springer

【 摘 要 】

BackgroundA key challenge in analyzing high throughput Single Nucleotide Polymorphism (SNP) arrays is the accurate inference of genotypes for SNPs with low minor allele frequencies. A number of calling algorithms have been developed to infer genotypes for common SNPs, but they are limited in their performance in calling rare SNPs. The existing algorithms can be broadly classified into three categories, including: population-based methods, SNP-based methods, and a hybrid of the two approaches. Despite the relatively better performance of the hybrid approach, it is still challenging to analyze rare SNPs.ResultsWe propose to utilize information from samples with known genotypes to develop a two stage genotyping procedure, namely M3-S, for rare SNP calling. This new approach can improve genotyping accuracy through clearly defining the boundaries of genotype clusters from samples with known genotypes, and enlarge the call rate by combining the simulated data based on the inferred genotype clusters information with the study population.ConclusionsApplications to real data demonstrates that this new approach M3-S outperforms existing methods in calling rare SNPs.

【 授权许可】

CC BY   
© Li and Zhao. 2015