doi: 10.1371/journal.pgen.1001131.
A novel statistic for genome-wide interaction analysis
Affiliations
- PMID: 20885795
- PMCID: PMC2944798
- DOI: 10.1371/journal.pgen.1001131
Item in Clipboard
A novel statistic for genome-wide interaction analysis
PLoS Genet.
.
Abstract
Although great progress in genome-wide association studies (GWAS) has been made, the significant SNP associations identified by GWAS account for only a few percent of the genetic variance, leading many to question where and how we can find the missing heritability. There is increasing interest in genome-wide interaction analysis as a possible source of finding heritability unexplained by current GWAS. However, the existing statistics for testing interaction have low power for genome-wide interaction analysis. To meet challenges raised by genome-wide interactional analysis, we have developed a novel statistic for testing interaction between two loci (either linked or unlinked). The null distribution and the type I error rates of the new statistic for testing interaction are validated using simulations. Extensive power studies show that the developed statistic has much higher power to detect interaction than classical logistic regression. The results identified 44 and 211 pairs of SNPs showing significant evidence of interactions with FDR<0.001 and 0.001<FDR<0.003, respectively, which were seen in two independent studies of psoriasis. These included five interacting pairs of SNPs in genes LST1/NCR3, CXCR5/BCL9L, and GLS2, some of which were located in the target sites of miR-324-3p, miR-433, and miR-382, as well as 15 pairs of interacting SNPs that had nonsynonymous substitutions. Our results demonstrated that genome-wide interaction analysis is a valuable tool for finding remaining missing heritability unexplained by the current GWAS, and the developed novel statistic is able to search significant interaction between SNPs across the genome. Real data analysis showed that the results of genome-wide interaction analysis can be replicated in two independent studies.
Conflict of interest statement
The authors have declared that no competing interests exist.
Figures
(A) Quantile-quantile plots for the test statistic
in dataset
1. The P-values (<
) for the
test are plotted (as −log10 values) as a function of its expected
p values. (B) Quantile-quantile plots for the test statistic
in dataset
2. The P-values (<
) for the
test are plotted (as −log10 values) as a function of its expected
p values.
in dataset
1. The P-values (<) for the
test are plotted (as −log10 values) as a function of its expected
p values. (B) Quantile-quantile plots for the test statistic
in dataset
2. The P-values (<) for the
test are plotted (as −log10 values) as a function of its expected
p values.
(A) The power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus recessive
recessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.05, and the
odds-ratios at two loci were 
. (B) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus recessive
recessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.01, and the
odds-ratios at two loci were 
. (C) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus recessive
recessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.001, and the
odds-ratios at two loci were 
.
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus recessiverecessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.05, and the
odds-ratios at two loci were . (B) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus recessiverecessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.01, and the
odds-ratios at two loci were . (C) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus recessiverecessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.001, and the
odds-ratios at two loci were .
(A) The power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus dominant
dominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.05, and the
odds-ratios at two loci were 
. (B) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus dominant
dominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.01, and the
odds-ratios at two loci were 
. (C) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio 
under a
two-locus dominant
dominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were 
.
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus dominantdominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.05, and the
odds-ratios at two loci were . (B) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus dominantdominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.01, and the
odds-ratios at two loci were . (C) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two linked loci as a function of
traditional odds-ratio under a
two-locus dominantdominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were .
(A) The power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio 
under a
two-locus additive
additive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.05, and the
odds-ratios at two loci were 
. (B) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio 
under a
two-locus additive
additive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.01, and the
odds-ratios at two loci were 
. (C) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio 
under a
two-locus additive
additive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were 
.
, the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio under a
two-locus additiveadditive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.05, and the
odds-ratios at two loci were . (B) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio under a
two-locus additiveadditive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.01, and the
odds-ratios at two loci were . (C) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression for
testing interaction between two linked loci analysis as a function of
traditional odds-ratio under a
two-locus additiveadditive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were .
(A) The power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio 
under a
two-locus recessive
recessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.001, and the
odds-ratios at two loci were 
. (B) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio 
under a
two-locus dominant
dominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were 
. (C) The
power of the test statistic 
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio 
under a
two-locus additive
additive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were 
.
, the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio under a
two-locus recessiverecessive
disease model, where the number of individuals in both the case and
control groups is 2,000, the significance level is 0.001, and the
odds-ratios at two loci were . (B) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio under a
two-locus dominantdominant
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were . (C) The
power of the test statistic , the
“fast-epistasis” in PLINK and logistic regression analysis
for testing interaction between two unlinked loci as a function of
traditional odds-ratio under a
two-locus additiveadditive
disease model, where the number of individuals in both the case and
control groups is 1,000, the significance level is 0.001, and the
odds-ratios at two loci were .
Each pathway was represented by an ellipse with the number. The SNPs were
represented by nodes and placed insight their located pathways. Nearby
each SNP there was its RS number and the name of its located gene. The
pathway and its harbored SNPs were labeled by the same color. The
interacting SNPs were connected by the solid light green lines.
Comment in
-
Improved statistics for genome-wide interaction analysis.PLoS Genet. 2012;8(4):e1002625. doi: 10.1371/journal.pgen.1002625. Epub 2012 Apr 5. PLoS Genet. 2012. PMID: 22496670 Free PMC article.
Similar articles
-
Gene, pathway and network frameworks to identify epistatic interactions of single nucleotide polymorphisms derived from GWAS data.BMC Syst Biol. 2012;6 Suppl 3(Suppl 3):S15. doi: 10.1186/1752-0509-6-S3-S15. Epub 2012 Dec 17. BMC Syst Biol. 2012. PMID: 23281810 Free PMC article.
-
Test for interaction between two unlinked loci.Am J Hum Genet. 2006 Nov;79(5):831-45. doi: 10.1086/508571. Epub 2006 Sep 21. Am J Hum Genet. 2006. PMID: 17033960 Free PMC article.
-
Testing gene-gene interactions in genome wide association studies.Genet Epidemiol. 2014 Feb;38(2):123-34. doi: 10.1002/gepi.21786. Epub 2014 Jan 15. Genet Epidemiol. 2014. PMID: 24431225 Free PMC article.
-
A review of genome-wide association studies for multiple sclerosis: classical and hypothesis-driven approaches.Hum Genet. 2015 Nov;134(11-12):1143-62. doi: 10.1007/s00439-015-1601-2. Epub 2015 Sep 25. Hum Genet. 2015. PMID: 26407970 Review.
-
Gene--environment-wide association studies: emerging approaches.Nat Rev Genet. 2010 Apr;11(4):259-72. doi: 10.1038/nrg2764. Nat Rev Genet. 2010. PMID: 20212493 Free PMC article. Review.
Cited by
-
Functional Regression Models for Epistasis Analysis of Multiple Quantitative Traits.PLoS Genet. 2016 Apr 22;12(4):e1005965. doi: 10.1371/journal.pgen.1005965. eCollection 2016 Apr. PLoS Genet. 2016. PMID: 27104857 Free PMC article.
-
Gene-Based Nonparametric Testing of Interactions Using Distance Correlation Coefficient in Case-Control Association Studies.Genes (Basel). 2018 Dec 5;9(12):608. doi: 10.3390/genes9120608. Genes (Basel). 2018. PMID: 30563156 Free PMC article.
-
Genome-wide eQTLs and heritability for gene expression traits in unrelated individuals.BMC Genomics. 2014 Jan 9;15(1):13. doi: 10.1186/1471-2164-15-13. BMC Genomics. 2014. PMID: 24405759 Free PMC article.
-
Genotypic variability-based genome-wide association study identifies non-additive loci HLA-C and IL12B for psoriasis.J Hum Genet. 2018 Mar;63(3):289-296. doi: 10.1038/s10038-017-0350-6. Epub 2017 Dec 19. J Hum Genet. 2018. PMID: 29259305
-
Detecting epistasis in human complex traits.Nat Rev Genet. 2014 Nov;15(11):722-33. doi: 10.1038/nrg3747. Epub 2014 Sep 9. Nat Rev Genet. 2014. PMID: 25200660 Review.
References
Publication types
MeSH terms
Substances
Grants and funding
LinkOut - more resources
Full Text Sources
