Recent studies have uncovered a strong effect of host genetic variation on the composition of host-associated microbiota. Here, we present HOMINID, a computational approach based on Lasso linear regression, that given host genetic variation and microbiome taxonomic composition data, identifies host single nucleotide polymorphisms (SNPs) that are correlated with microbial taxa abundances. Using simulated data, we show that HOMINID has accuracy in identifying associated SNPs and performs better compared with existing methods. We also show that HOMINID can accurately identify the microbial taxa that are correlated with associated SNPs. Lastly, by using HOMINID on real data of human genetic variation and microbiome composition, we identified 13 human SNPs in which genetic variation is correlated with microbiome taxonomic composition across body sites. In conclusion, HOMINID is a powerful method to detect host genetic variants linked to microbiome composition and can facilitate discovery of mechanisms controlling host-microbiome interactions.
Bibliographical noteFunding Information:
This work is supported in part by funds from the University of Minnesota College of Biological Sciences, The Randy Shaver Cancer Research and Community Fund, Institutional Research Grant No. 124166-IRG-58–001-55-IRG53 from the American Cancer Society, and a Research Fellowship from The Alfred P. Sloan Foundation. This work was facilitated in part by computational resources provided by the Minnesota Supercomputing Institute.
© The Author(s) 2017.
- Host genetics
- Machine learning