Causal graph-based analysis of genome-wide association data in rheumatoid arthritis

Abstract

BackgroundGWAS owe their popularity to the expectation that they will make a major impact on diagnosis, prognosis and management of disease by uncovering genetics underlying clinical phenotypes. The dominant paradigm in GWAS data analysis so far consists of extensive reliance on methods that emphasize contribution of individual SNPs to statistical association with phenotypes. Multivariate methods, however, can extract more information by considering associations of multiple SNPs simultaneously. Recent advances in other genomics domains pinpoint multivariate causal graph-based inference as a promising principled analysis framework for high-throughput data. Designed to discover biomarkers in the local causal pathway of the phenotype, these methods lead to accurate and highly parsimonious multivariate predictive models. In this paper, we investigate the applicability of causal graph-based method TIE* to analysis of GWAS data. To test the utility of TIE*, we focus on anti-CCP positive rheumatoid arthritis (RA) GWAS datasets, where there is a general consensus in the community about the major genetic determinants of the disease.ResultsApplication of TIE* to the North American Rheumatoid Arthritis Cohort (NARAC) GWAS data results in six SNPs, mostly from the MHC locus. Using these SNPs we develop two predictive models that can classify cases and disease-free controls with an accuracy of 0.81 area under the ROC curve, as verified in independent testing data from the same cohort. The predictive performance of these models generalizes reasonably well to Swedish subjects from the closely related but not identical Epidemiological Investigation of Rheumatoid Arthritis (EIRA) cohort with 0.71-0.78 area under the ROC curve. Moreover, the SNPs identified by the TIE* method render many other previously known SNP associations conditionally independent of the phenotype.ConclusionsOur experiments demonstrate that application of TIE* captures maximum amount of genetic information about RA in the data and recapitulates the major consensus findings about the genetic factors of this disease. In addition, TIE* yields reproducible markers and signatures of RA. This suggests that principled multivariate causal and predictive framework for GWAS analysis empowers the community with a new tool for high-quality and more efficient discovery.ReviewersThis article was reviewed by Prof. Anthony Almudevar, Dr. Eugene V. Koonin, and Prof. Marianthi Markatou.

Genome-wide association study identified the human leukocyte antigen region as a novel locus for plasma beta-2 microglobulin

“…Given the HLA region have been implicated in many complex diseases (Alekseyenko et al 2011; Fernando et al 2012; International Multiple Sclerosis Genetics et al 2011; Lucena et al 2011; McLachlan et al 2011; Othman et al 2011; Ramos et al 2011; Skinningsrud et al 2011), we tested for the associations between the six B2M index SNPs at the HLA locus and all-cause mortality in the ARIC study and found no evidence of associations (p>8.3×10 −3 =0.05/6, Supplementary Table 5). …”

A survey about methods dedicated to epistasis detection

“…A different but not less appealing Bayesian strategy is the Markov blanket-based method. It allows discovery of SNPs in the local pathway of the phenotype, also referred to as “local causal SNPs” (Alekseyenko et al, 2011 ). In the context of GWAS, this strategy is used to avoid the time-consuming training processes like tree-growing of random forests or structure learning of a full Bayesian network.…”

“…A SNP Y is detected as a false positive if it is independent of the phenotype given a SNP subset of canMB . Three implementations of this approach were recently developed: DASSO-MB (Han et al, 2010 ), TIE * (Alekseyenko et al, 2011 ; Statnikov et al, 2013 ) and IMBED (Yanlan and Jiawei, 2012 ), and all proved to be more sample-efficient than BEAM, i.e., less samples are needed to reach the same power of detection as BEAM. In DASSO-MB (Han et al, 2010 , Han and coworkers postulate that, in epistatic interaction studies, only causal SNPs are sought, and consequently only parent nodes of the phenotype have to be detected.…”

“…Hence, DASSO-MB represents a more specific application of the Markov Blanket approach. Considering a set of 19 SNPs already known to be associated with rheumatoid arthritis, an application of TIE * (Target Information Equivalency) showed that a Markov blanket-based approach could make the whole SNP set independent of the phenotype when conditioned on three other SNPs identified in the Markov blanket (Alekseyenko et al, 2011 ). In other words, the reported SNP set does not provide any predictive information about the disease status beyond that brought by the three SNPs identified with the Markov blanket.…”

Trends and Developments in Bioinformatics in 2010: Prospects and Perspectives

“…Multivariate vs Univariate analyses for GWAS studies. The year 2010 continued a trend seen in 2009, witnessing the emergence of multivariate analysis of GWAS data [34][35][36][37]. Although more than 1,000 GWAS studies have been completed and corresponding datasets made available [38] (http://www.genome.gov/ gwastudies/) the dominant analysis paradigm was univariate (one SNP at a time).…”