Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps

Abstract

We aggregated genome-wide genotyping data from 32 European-descent GWAS (74,124 T2D cases, 824,006 controls) imputed to high-density reference panels of >30,000 sequenced haplotypes. Analysis of ˜27M variants (˜21M with minor allele frequency [MAF]<5%), identified 243 genome-wide significant loci (p<5x10-8; MAF 0.02%-50%; odds ratio [OR] 1.04-8.05), 135 not previously-implicated in T2D-predisposition. Conditional analyses revealed 160 additional distinct association signals (p<10-5) within the identified loci. The combined set of 403 T2D-risk signals includes 56 low-frequency (0.5%≤MAF<5%) and 24 rare (MAF<0.5%) index SNPs at 60 loci, including 14 with estimated allelic OR>2. Forty-one of the signals displayed effect-size heterogeneity between BMI-unadjusted and adjusted analyses. Increased sample size and improved imputation led to substantially more precise localisation of causal variants than previously attained: at 51 signals, the lead variant after fine-mapping accounted for >80% posterior probability of association (PPA) and at 18 of these, PPA exceeded 99%. Integration with islet regulatory annotations enriched for T2D association further reduced median credible set size (from 42 variants to 32) and extended the number of index variants with PPA>80% to 73. Although most signals mapped to regulatory sequence, we identified 18 genes as human validated therapeutic targets through coding variants that are causal for disease. Genome wide chip heritability accounted for 18% of T2D-risk, and individuals in the 2.5% extremes of a polygenic risk score generated from the GWAS data differed >9-fold in risk. Our observations highlight how increases in sample size and variant diversity deliver enhanced discovery and single-variant resolution of causal T2D-risk alleles, and the consequent impact on mechanistic insights and clinical translation.

Altered N‐glycosylation profiles as potential biomarkers and drug targets in diabetes

“…ST6GAL1 encodes an N‐glycosylation pathway protein responsible for the transfer of α2,6‐linked sialic acid to galactose‐containing substrates . Identified type 2 diabetes risk allele in European‐descent individuals was associated with an increase in ST6GAL1 expression in islets . This finding further corroborates the hypothesis that highly branched N‐glycans (which are also highly sialylated) are associated with higher risk of developing type 2 diabetes .…”

Transethnic meta‐analysis of metabolic syndrome in a multiethnic study

“…This is because TransMeta would produce slightly lower p values than the other methods at an SNP, leading to a suggestive result by TransMeta while the other method's p values were only on the order of 10 −6 or 10 −5 . While this locus was primarily driven by the MA sample in the GENNID sample, this locus showed evidence of replication in the UK BioBank (UKBB; Mahajan et al, 2018), with the lead SNP at this locus being associated with weight in the UKBB with a p = 1.88e−4. Since TransMeta contains FE in its hierarchical framework, TransMeta detected most of the loci detected by FE; there was also a large overlap between the loci detected by RE2 and TransMeta (Figure 2).…”

On the differences between mega‐ and meta‐imputation and analysis exemplified on the genetics of age‐related macular degeneration

“…It is also possible that models including more study‐specific covariates will have more issues with rare variants than the model with the fewest covariates (Model I). There are different GWAMA approaches of how to filter variants with association results from logistic regression to ensure reliable estimates; we defined a variant as “analyzable” for a specific imputation/model (i.e., yielding trustworthy association test‐statistics), if it had a MAC > 5, nonmissing effects/standard errors ( SE )/ P ‐values, and SE < 10 across all studies combined (for mega‐models, Models I–IV) or per study (for meta‐model, Model V), in analogy as described before (Mahajan et al, ). A higher number of analyzable variants in one imputation/model combination compared to others can be considered an advantage.…”