OSCA: a tool for omic-data-based complex trait analysis

Futao Zhang,Peter M Visscher,Ting Qi,Qian Zhang,Jian Yang,Marta F Nabais,Naomi R Wray,Wenhan Chen,Zhihong Zhu,Allan F Mcrae,Ian J Deary

doi:10.1186/s13059-019-1718-z

Futao Zhang, Peter M Visscher + Show 9 more

Open Access

https://doi.org/10.1186/s13059-019-1718-z

Copy DOI

Abstract

The rapid increase of omic data has greatly facilitated the investigation of associations between omic profiles such as DNA methylation (DNAm) and complex traits in large cohorts. Here, we propose a mixed-linear-model-based method called MOMENT that tests for association between a DNAm probe and trait with all other distal probes fitted in multiple random-effect components to account for unobserved confounders. We demonstrate by simulations that MOMENT shows a lower false positive rate and more robustness than existing methods. MOMENT has been implemented in a versatile software package called OSCA together with a number of other implementations for omic-data-based analyses.

Highlights

The rapid proliferation of genetic and omic data in large cohort-based samples in the past decade has greatly advanced our understanding of the genetic architecture of omic profiles and the molecular mechanisms underpinning the genetic variation of human complex traits [1,2,3]
Overview of the OSCA software OSCA comprises four main modules: (1) data management for which we designed a binary format to efficiently store and manage omic data; (2) linear-regression- and mixed linear model (MLM)-based methods to test for associations between omic measures and complex traits; (3) methods to estimate the proportion of variance in a complex trait captured by all the measures of one or multiple omic profiles and to predict the trait phenotype in a new sample based on the joint effects of all omic measures estimated in a discovery sample; and (4) an efficient implementation of the methods to identify genetic variants associated with an omic profile, e.g., DNA methylation quantitative trait loci analysis
MLM-based omic association analysis methods One of the primary applications of OSCA is to test for associations between omic measures (e.g., DNA methylation (DNAm) probes) and a complex trait (e.g., body mass index (BMI)) correcting for confounding effects

Summary

Introduction

The rapid proliferation of genetic and omic data in large cohort-based samples in the past decade has greatly advanced our understanding of the genetic architecture of omic profiles and the molecular mechanisms underpinning the genetic variation of human complex traits [1,2,3] These advances include the identification of a large number of genetic variants associated with gene expression [4, 5], DNA methylation [6, 7], histone modification [8, 9], and protein abundance [10, 11]; the discovery of omic measures associated with complex traits [12, 13]; the improved accuracy in predicting a trait using omic data [14, 15]; and the prioritization of gene targets for complex traits by integrating genetic and omic data in large samples [3, 13, 16,17,18]. Uncharacterized confounders with small to moderate effects and numerous correlations between distal DNAm probes (e.g., those on different chromosomes) induced by the confounders may not be well captured by either a fixed number of principal features or a subset of selected probes

Methods

Results

Conclusion