Limits...
GPA: a statistical approach to prioritizing GWAS results by integrating pleiotropy and annotation.

Chung D, Yang C, Li C, Gelernter J, Zhao H - PLoS Genet. (2014)

Bottom Line: Results from Genome-Wide Association Studies (GWAS) have shown that complex diseases are often affected by many genetic variants with small or moderate effects.Using our hypothesis testing framework, statistically significant pleiotropic effects were detected among these psychiatric disorders, and the markers annotated in the central nervous system genes and eQTLs from the Genotype-Tissue Expression (GTEx) database were significantly enriched.GPA was able to detect cell lines that are biologically more relevant to bladder cancer.

View Article: PubMed Central - PubMed

Affiliation: Department of Biostatistics, Yale School of Public Health, New Haven, Connecticut, United States of America; Department of Public Health Sciences, Medical University of South Carolina, Charleston, South Carolina, United States of America.

ABSTRACT
Results from Genome-Wide Association Studies (GWAS) have shown that complex diseases are often affected by many genetic variants with small or moderate effects. Identifications of these risk variants remain a very challenging problem. There is a need to develop more powerful statistical methods to leverage available information to improve upon traditional approaches that focus on a single GWAS dataset without incorporating additional data. In this paper, we propose a novel statistical approach, GPA (Genetic analysis incorporating Pleiotropy and Annotation), to increase statistical power to identify risk variants through joint analysis of multiple GWAS data sets and annotation information because: (1) accumulating evidence suggests that different complex diseases share common risk bases, i.e., pleiotropy; and (2) functionally annotated variants have been consistently demonstrated to be enriched among GWAS hits. GPA can integrate multiple GWAS datasets and functional annotations to seek association signals, and it can also perform hypothesis testing to test the presence of pleiotropy and enrichment of functional annotation. Statistical inference of the model parameters and SNP ranking is achieved through an EM algorithm that can handle genome-wide markers efficiently. When we applied GPA to jointly analyze five psychiatric disorders with annotation information, not only did GPA identify many weak signals missed by the traditional single phenotype analysis, but it also revealed relationships in the genetic architecture of these disorders. Using our hypothesis testing framework, statistically significant pleiotropic effects were detected among these psychiatric disorders, and the markers annotated in the central nervous system genes and eQTLs from the Genotype-Tissue Expression (GTEx) database were significantly enriched. We also applied GPA to a bladder cancer GWAS data set with the ENCODE DNase-seq data from 125 cell lines. GPA was able to detect cell lines that are biologically more relevant to bladder cancer. The R implementation of GPA is currently available at http://dongjunchung.github.io/GPA/.

Show MeSH

Related in: MedlinePlus

Global false discovery rates of GPA at sample size  = 5000 and number of risk SNPs  = 1000.Upper panel: Global false discovery rates of GPA with annotation. Lower panel: Global false discovery rates of GPA without annotation. From left to right: FDR of first GWAS (joint analysis), FDR of second GWAS (joint analysis), FDR of first GWAS (separate analysis), FDR of second GWAS (separate analysis) and FDR of risk variants shared by both GWAS. For all scenarios, the global false discovery rates of GPA are controlled at the nominal level.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4230845&req=5

pgen-1004787-g002: Global false discovery rates of GPA at sample size  = 5000 and number of risk SNPs  = 1000.Upper panel: Global false discovery rates of GPA with annotation. Lower panel: Global false discovery rates of GPA without annotation. From left to right: FDR of first GWAS (joint analysis), FDR of second GWAS (joint analysis), FDR of first GWAS (separate analysis), FDR of second GWAS (separate analysis) and FDR of risk variants shared by both GWAS. For all scenarios, the global false discovery rates of GPA are controlled at the nominal level.

Mentions: We first evaluated SNP prioritization performance of GPA. Specifically, after the two data sets were simulated, we obtained the p-value for each SNP in each disease using a test with one degree of freedom. Then we analyzed the simulated data using our GPA method in the following four modes: 1. analyzing the two diseases separately without the annotation data; 2. analyzing the two diseases separately with the annotation data; 3. analyzing the two diseases jointly without the annotation data; and 4. analyzing the two diseases jointly with the annotation data. In each mode, we compared the order of the local FDR obtained using GPA against the actual risk status of the SNPs to calculate the area under the receiver operating characteristic curve (AUC) as a measure of risk SNP prioritization accuracy. The left panel of Figure 1 shows the AUCs from the four modes with and (results for other scenarios are shown in Figures S10-S20 in Text S1). Because all the simulation parameters were the same for the two diseases, only the results for the first disease are shown. We can see that incorporating either annotation information or pleiotropy between the two diseases improved the prioritization performance. In particular, as the proportion of shared risk SNPs increased, the prioritization performance also improved. Given the local FDR obtained using GPA, we controlled the global FDR at 0.2 and calculated the average power to identify the true risk SNPs. GPA performance measured by partial AUC and power for and are shown in the middle and right panels of Figure 1, respectively (results for other scenarios are shown in Figures S10–S20 in Text S1). We also evaluated the actual FDR and found that the FDR was indeed controlled at 0.2 with occasional slight conservativeness (Figure 2 and Figures S21–S31 in Text S1). In addition, we evaluated the actual FDR in the presence of linkage disequilibrium (LD) among SNPs. The details of the simulations and results are given in Figure S1 in Text S1. These results suggest that GPA can improve the power of identifying risk variants while controling FDR at the nominal level despite the mismatch between GPA and the generative model in our simulation study.


GPA: a statistical approach to prioritizing GWAS results by integrating pleiotropy and annotation.

Chung D, Yang C, Li C, Gelernter J, Zhao H - PLoS Genet. (2014)

Global false discovery rates of GPA at sample size  = 5000 and number of risk SNPs  = 1000.Upper panel: Global false discovery rates of GPA with annotation. Lower panel: Global false discovery rates of GPA without annotation. From left to right: FDR of first GWAS (joint analysis), FDR of second GWAS (joint analysis), FDR of first GWAS (separate analysis), FDR of second GWAS (separate analysis) and FDR of risk variants shared by both GWAS. For all scenarios, the global false discovery rates of GPA are controlled at the nominal level.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4230845&req=5

pgen-1004787-g002: Global false discovery rates of GPA at sample size  = 5000 and number of risk SNPs  = 1000.Upper panel: Global false discovery rates of GPA with annotation. Lower panel: Global false discovery rates of GPA without annotation. From left to right: FDR of first GWAS (joint analysis), FDR of second GWAS (joint analysis), FDR of first GWAS (separate analysis), FDR of second GWAS (separate analysis) and FDR of risk variants shared by both GWAS. For all scenarios, the global false discovery rates of GPA are controlled at the nominal level.
Mentions: We first evaluated SNP prioritization performance of GPA. Specifically, after the two data sets were simulated, we obtained the p-value for each SNP in each disease using a test with one degree of freedom. Then we analyzed the simulated data using our GPA method in the following four modes: 1. analyzing the two diseases separately without the annotation data; 2. analyzing the two diseases separately with the annotation data; 3. analyzing the two diseases jointly without the annotation data; and 4. analyzing the two diseases jointly with the annotation data. In each mode, we compared the order of the local FDR obtained using GPA against the actual risk status of the SNPs to calculate the area under the receiver operating characteristic curve (AUC) as a measure of risk SNP prioritization accuracy. The left panel of Figure 1 shows the AUCs from the four modes with and (results for other scenarios are shown in Figures S10-S20 in Text S1). Because all the simulation parameters were the same for the two diseases, only the results for the first disease are shown. We can see that incorporating either annotation information or pleiotropy between the two diseases improved the prioritization performance. In particular, as the proportion of shared risk SNPs increased, the prioritization performance also improved. Given the local FDR obtained using GPA, we controlled the global FDR at 0.2 and calculated the average power to identify the true risk SNPs. GPA performance measured by partial AUC and power for and are shown in the middle and right panels of Figure 1, respectively (results for other scenarios are shown in Figures S10–S20 in Text S1). We also evaluated the actual FDR and found that the FDR was indeed controlled at 0.2 with occasional slight conservativeness (Figure 2 and Figures S21–S31 in Text S1). In addition, we evaluated the actual FDR in the presence of linkage disequilibrium (LD) among SNPs. The details of the simulations and results are given in Figure S1 in Text S1. These results suggest that GPA can improve the power of identifying risk variants while controling FDR at the nominal level despite the mismatch between GPA and the generative model in our simulation study.

Bottom Line: Results from Genome-Wide Association Studies (GWAS) have shown that complex diseases are often affected by many genetic variants with small or moderate effects.Using our hypothesis testing framework, statistically significant pleiotropic effects were detected among these psychiatric disorders, and the markers annotated in the central nervous system genes and eQTLs from the Genotype-Tissue Expression (GTEx) database were significantly enriched.GPA was able to detect cell lines that are biologically more relevant to bladder cancer.

View Article: PubMed Central - PubMed

Affiliation: Department of Biostatistics, Yale School of Public Health, New Haven, Connecticut, United States of America; Department of Public Health Sciences, Medical University of South Carolina, Charleston, South Carolina, United States of America.

ABSTRACT
Results from Genome-Wide Association Studies (GWAS) have shown that complex diseases are often affected by many genetic variants with small or moderate effects. Identifications of these risk variants remain a very challenging problem. There is a need to develop more powerful statistical methods to leverage available information to improve upon traditional approaches that focus on a single GWAS dataset without incorporating additional data. In this paper, we propose a novel statistical approach, GPA (Genetic analysis incorporating Pleiotropy and Annotation), to increase statistical power to identify risk variants through joint analysis of multiple GWAS data sets and annotation information because: (1) accumulating evidence suggests that different complex diseases share common risk bases, i.e., pleiotropy; and (2) functionally annotated variants have been consistently demonstrated to be enriched among GWAS hits. GPA can integrate multiple GWAS datasets and functional annotations to seek association signals, and it can also perform hypothesis testing to test the presence of pleiotropy and enrichment of functional annotation. Statistical inference of the model parameters and SNP ranking is achieved through an EM algorithm that can handle genome-wide markers efficiently. When we applied GPA to jointly analyze five psychiatric disorders with annotation information, not only did GPA identify many weak signals missed by the traditional single phenotype analysis, but it also revealed relationships in the genetic architecture of these disorders. Using our hypothesis testing framework, statistically significant pleiotropic effects were detected among these psychiatric disorders, and the markers annotated in the central nervous system genes and eQTLs from the Genotype-Tissue Expression (GTEx) database were significantly enriched. We also applied GPA to a bladder cancer GWAS data set with the ENCODE DNase-seq data from 125 cell lines. GPA was able to detect cell lines that are biologically more relevant to bladder cancer. The R implementation of GPA is currently available at http://dongjunchung.github.io/GPA/.

Show MeSH
Related in: MedlinePlus