Analyzing and Reconciling Colocalization and Transcriptome-wide Association Studies from the Perspective of Inferential Reproducibility

Abstract

Transcriptome-wide association studies and colocalization analysis are popular computational approaches for integrating genetic association data from molecular and complex traits. They show the unique ability to go beyond variant-level genetic association evidence and implicate critical functional units, e.g., genes, in disease etiology. However, in practice, when the two approaches are applied to the same molecular and complex trait data, the inference results can be markedly different. This paper systematically investigates the inferential reproducibility between the two approaches through theoretical derivation, numerical experiments, and analyses of 4 complex trait GWAS and GTEx eQTL data. We identify two classes of inconsistent inference results. We find that the first class of inconsistent results may suggest an interesting biological phenomenon, i.e., horizontal pleiotropy; thus, the two approaches are truly complementary. The inconsistency in the second class can be understood and effectively reconciled. To this end, we propose a novel approach for locus-level colocalization analysis. We demonstrate that the joint TWAS and locus-level colocalization analysis improves specificity and sensitivity for implicating biological-relevant genes.

Mapping genomic regulation of kidney disease and traits through high-resolution and interpretable eQTLs

“…For UACR at an FDR ≤ 5%, we identified 137 significant gene-trait pairs in GLOM (Figure 5B) and 179 in TUBE (Table S7, Figures S8A-C). We also found a significant correlation (ρ = 0.57, P ≤ 2.2×10 -16 ) between colocalization and PTWAS signals (Figure 5C, Figure S8D-F), demonstrating the consistency of inference results when different analytical approaches are applied to the same dataset (Hukku et al, 2021a).…”

Probabilistic integration of transcriptome-wide association studies and colocalization analysis prioritizes molecular pathways of complex traits

“…In the analysis presented in this paper, we use the gene locus-level colocalization probabilities (GLCPs) from fastENLOC as input for INTACT, unless otherwise stated. GLCPs have the advantage of improving the power of colocalization analysis by accommodating the inaccuracy of pinpointing causal variants from genetic association analysis [27]. A 20 valid monotonic mapping for INTACT, f , should satisfy properties (3) and (4).…”

“…We first compare INTACT's performance as a method for inferring PCGs to some existing computational approaches. The methods for comparison roughly fall into three categories: colocalization-focused approaches represented by fastENLOC [10,27] with two types of genelevel quantification (GLCP and GRCP); TWAS-only approaches represented by PTWAS and an 10 implementation of LDA MR-Egger/FOCUS algorithm [14,9]; and the post hoc joint TWAS and colocalization analysis method described in Hukku et al [27,10]. For each dataset consisting of 1198 genes, we set the target FDR control level at 5% for all methods.…”

“…However, resolving within-gene LD hitchhiking remains an open problem [14]. Hukku et al [27] explore the inferential reproducibility of PCG implications between TWAS and colocalization analysis, concluding that these two procedures can be complementary. Particularly, they argue that both TWAS and colocalization signals should be present for a putative causal gene based on theoretical and empirical evidence.…”

“…We apply INTACT to re-analyze the GWAS data from 4 complex traits (HDL and LDL from the GLGC consortium; standing height from the UK Biobank; coronary artery disease from the CARDioGRAM consortium) and the eQTL data across 49 tissues from the GTEx project (v8). Hukku et al[27] report the integrative analysis results of this data set using the post hoc approach. The comparison between INTACT and the original Hukku et al analysis is summarized in Supplemental TableS4.…”