Skip to content

HyPrColoc

Summary

HyPrColoc (Hypothesis Prioritisation for multi-trait Colocalization) is an efficient deterministic Bayesian algorithm designed for multi-trait genetic colocalization analyses. It uses GWAS summary statistics to simultaneously identify shared causal variants across large numbers of traits, scaling to dozens or hundreds of traits in seconds. Developed by the Burgess Group, it features a branch-and-bound divisive clustering algorithm to identify distinct subsets of traits that colocalize at different causal variants in a genomic region.

Methodology

In genetic fine-mapping and colocalization, multi-trait analyses increase statistical power to prioritize causal variants. However, standard multi-trait colocalization methods like MOLOC suffer from exponential computational complexity, scaling as $O((Q+1)^m)$ where $Q$ is the number of variants and $m$ is the number of traits.

HyPrColoc overcomes this bottleneck by approximating the Posterior Probability of Full Colocalization (PPFC) using only a subset of configurations. The method computes: 1. Regional Association Probability ($P_R$): The probability that all traits share an association with one or more variants within the region. 2. Alignment Probability ($P_A$): The probability that the shared associations are due to a single shared causal variant.

The overall posterior probability is approximated as: $$\widehat{\text{PPFC}} = P_R P_A$$ This formulation reduces the computational complexity to $O(mQ^2)$ (as of 2021-02-03), making it feasible to perform genome-wide colocalization analyses across vast numbers of molecular, metabolic, and clinical traits.

Applications and Capabilities

  • Divisive Clustering: When traits do not colocalize, HyPrColoc uses a divisive branch-and-bound algorithm to partition the traits into clusters of colocalizing traits.
  • eQTL and pQTL Integration: In its validation study, HyPrColoc was used to integrate coronary heart disease (CHD) GWAS with whole blood gene expression (eQTL) and protein expression (pQTL) data, identifying 43 colocalized regions (including 5 novel CHD loci) and prioritizing causal genes such as GUCY1A1 and BCAR1.
  • Sensitivity Analysis: HyPrColoc includes tools to visualize changes in identified clusters across different priors ($p_c$) and thresholds, helping to diagnose potential false positives.

Citations

  • Foley, C. N., Staley, J. R., Breen, P. G., Sun, B. B., Kirk, P. D. W., Burgess, S., & Howson, J. M. M. (2021). A fast and efficient colocalization algorithm for identifying shared genetic risk factors across multiple traits. Nature Communications, 12(1), 764. DOI: 10.1038/s41467-020-20885-8. Source paper: s41467-020-20885-8.pdf