Background: Fine-mapping is an analytical step for causal prioritization of the polymorphic variants in a trait-associated genomic region observed in genome-wide association studies (GWAS). Prioritization of causal variants can be challenging due to linkage disequilibrium (LD) patterns among hundreds to thousands of polymorphisms associated with a trait. Hence, we propose an ℓ0 graph norm shrinkage algorithm to disentangle LD patterns by dense LD blocks consisting of highly correlated single nucleotide polymorphisms (SNPs). We further incorporate the dense LD structure for fine-mapping. Based on graph theory, the concept of "dense" refers to a condition where a block is composed mainly of highly correlated SNPs. We demonstrated the application of our new fine-mapping method using a large UK Biobank (UKBB) sample related to nicotine addiction. We also evaluated and compared its performance with existing fine-mapping algorithms using simulations.
Results: Our results suggested that polymorphic variances in both neighboring and distant variants can be consolidated into dense blocks of highly correlated loci. Dense-LD outperformed comparable fine-mapping methods with increased sensitivity and reduced false-positive error rate for causal variant selection. Applying to a UKBB sample, this method replicated the loci reported in previous findings and suggested a strong association with nicotine addiction.
Conclusion: We found that the dense LD block structure can guide fine-mapping and accurately determine a parsimonious set of potential causal variants. Our approach is computationally efficient and allows fine-mapping of thousands of polymorphisms.