This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| COLOC | |
|---|---|
| Name | COLOC |
| Type | Statistical method |
| Field | Genetics; Genomics; Bioinformatics |
| First publication | 2014 |
| Primary authors | Giambartolomei et al. |
| Implementation | R; Python |
| License | Open source (varies) |
COLOC
COLOC is a Bayesian statistical method developed to test whether two association signals, typically from trait association studies and expression quantitative trait loci studies, share a common causal variant. It was introduced to integrate results from genome-wide association studies (Genome-wide association study) and molecular QTL mapping such as Expression quantitative trait locus studies to prioritize shared genetic mechanisms. The framework has been cited in analyses involving cohorts such as the 1000 Genomes Project, the UK Biobank, and disease consortia like the International Cancer Genome Consortium and the Psychiatric Genomics Consortium. COLOC has been used alongside resources like the GTEx Project, ENCODE Project, and the Roadmap Epigenomics Project to link loci implicated by studies including the CARDIoGRAM meta-analyses and the DIAGRAM consortium.
COLOC employs a Bayesian model comparison approach to evaluate five hypotheses about association status at a locus: no association with either trait, association with only the first trait, association with only the second trait, association with both traits but different causal variants, and association with both traits with a shared causal variant. The model uses summary statistics—effect sizes, standard errors, allele frequencies—from studies such as Wellcome Trust Case Control Consortium analyses or meta-analyses performed by METAL and leverages prior probabilities informed by studies like HapMap and 1000 Genomes Project. Likelihoods are computed assuming a single causal variant per locus; posterior probabilities are obtained via Bayes factors similar in spirit to methods used in fine-mapping approaches like CAVIAR and FINEMAP. COLOC can be applied to summary-level data from large projects such as GIANT (consortium), ENCODE Project, and trait-specific consortia including CARDIoGRAMplusC4D and ImmGen.
COLOC has been broadly applied to prioritize genes and mechanisms in studies of complex traits including cardiovascular disease examined by CARDIoGRAMplusC4D, metabolic traits from GIANT (consortium), autoimmune diseases studied by the Immunochip and the International Multiple Sclerosis Genetics Consortium, psychiatric disorders analyzed by the Psychiatric Genomics Consortium, and cancer susceptibility loci cataloged by The Cancer Genome Atlas. It is commonly used to link GWAS hits to regulatory variation mapped in GTEx Project tissues, enhancer annotations from ENCODE Project, and chromatin interaction maps from Hi-C and CHiA-PET datasets. COLOC has been incorporated into integrative pipelines combining annotations from CADD, ANNOVAR, and RegulomeDB to prioritize candidate causal genes in studies led by groups affiliated with institutions like Broad Institute and Wellcome Sanger Institute.
Primary implementations of the method are available as R packages originating from authors affiliated with academic centers; these packages accept summary statistics and provide posterior probability outputs compatible with visualization tools such as those from LocusZoom and plotting libraries used in workflows by Bioconductor. Variants and extensions have been integrated into platforms and tools developed by consortia including GTEx Project portals and analytical suites used at the European Bioinformatics Institute. Implementations interoperable with Python and workflow managers like Snakemake and Nextflow facilitate large-scale analyses in cloud environments provided by Amazon Web Services and Google Cloud Platform.
COLOC's single-causal-variant assumption per locus simplifies computations but may be violated in loci with allelic heterogeneity reported in studies such as those by International HapMap Consortium and analysis tools like GCTA. Its performance depends on accurate linkage disequilibrium estimates typically derived from reference panels like 1000 Genomes Project or population-specific cohorts such as UK Biobank; mismatches can bias posterior probabilities. Power is influenced by sample size disparities between datasets—examples include large GWAS meta-analyses versus smaller eQTL studies from GTEx Project—and by allele frequency differences highlighted in diverse-population projects like H3Africa. Alternatives and complementary approaches addressing multiple causal variants include methods such as eCAVIAR and SuSiE.
COLOC has featured in numerous high-impact studies linking GWAS loci to molecular traits: for example, integration of lipid trait GWAS from Global Lipids Genetics Consortium with liver eQTLs in GTEx Project implicated causal genes at loci interrogated by consortia like MIGen. In autoimmune disease research, applications combining data from the Immunochip and immune cell eQTLs cataloged by BLUEPRINT and ImmGen have nominated candidate genes influencing susceptibility. Psychiatric genomics efforts, leveraging results from the Psychiatric Genomics Consortium with brain-specific eQTLs and chromatin annotations from PsychENCODE, have used COLOC to prioritize targets at schizophrenia-associated loci. Cancer susceptibility analyses linking GWAS from Breast Cancer Association Consortium with tumor-specific expression datasets from The Cancer Genome Atlas have also employed COLOC-style inference.
Developments aim to relax the single-causal-variant assumption by integrating fine-mapping outputs from multi-variant methods like FINEMAP and SuSiE, incorporate cross-population LD modeling informed by projects such as Human Genome Diversity Project, and combine molecular QTL types including protein QTLs from pQTL studies and methylation QTLs from epigenetic consortia like Roadmap Epigenomics Project. Methodological enhancements may integrate functional priors from tools such as CADD, DeepSEA, and Eigen and exploit federated data frameworks used in initiatives like GA4GH to increase applicability across cohorts such as UK Biobank and regional biobanks. Ongoing software engineering efforts focus on scalable implementations compatible with cloud infrastructures provided by Amazon Web Services and Google Cloud Platform to support consortium-scale analyses.
Category:Statistical genetics