The landscape of GWAS validation; systematic review identifying 309 validated non-coding variants across 130 human diseases

Abstract

Background The remarkable growth of genome-wide association studies (GWAS) has created a critical need to experimentally validate the disease-associated variants, 90% of which involve non-coding variants. Methods To determine how the field is addressing this urgent need, we performed a comprehensive literature review identifying 36,676 articles. These were reduced to 1454 articles through a set of filters using natural language processing and ontology-based text-mining. This was followed by manual curation and cross-referencing against the GWAS catalog, yielding a final set of 286 articles. Results We identified 309 experimentally validated non-coding GWAS variants, regulating 252 genes across 130 human disease traits. These variants covered a variety of regulatory mechanisms. Interestingly, 70% (215/309) acted through cis-regulatory elements, with the remaining through promoters (22%, 70/309) or non-coding RNAs (8%, 24/309). Several validation approaches were utilized in these studies, including gene expression (n = 272), transcription factor binding (n = 175), reporter assays (n = 171), in vivo models (n = 104), genome editing (n = 96) and chromatin interaction (n = 33). Conclusions This review of the literature is the first to systematically evaluate the status and the landscape of experimentation being used to validate non-coding GWAS-identified variants. Our results clearly underscore the multifaceted approach needed for experimental validation, have practical implications on variant prioritization and considerations of target gene nomination. While the field has a long way to go to validate the thousands of GWAS associations, we show that progress is being made and provide exemplars of validation studies covering a wide variety of mechanisms, target genes, and disease areas.

From genome to phenome via the proteome: broad capture, antibody-based proteomics to explore disease mechanisms

“…The vast majority (~90%) of genetic variants identified in GWASs reside in non-coding regions of the genome (76), creating a challenge for variant-to-function annotation. In line with previous studies, we demonstrate the efficiency and ability of cis-pQTLs to prioritize causal candidate genes including reassignments at 40% of overlapping loci.…”

Machine Learning Reveals Genetic Modifiers of the Immune Microenvironment of Cancer

“…The information was mined from specific sections (depending on the concept) in the disease and gene monographs, as well as clinical synopses using IQVIA/Linguamatics I2E platform ([CSL STYLE ERROR: reference with no printed form.]). We further utilized internally built knowledge extraction pipelines and disambiguation routines (Alsheikh et al 2022). TDM queries were focused on paragraphs with clinical / human focus.…”

“…We further utilized internally built knowledge extraction pipelines and disambiguation routines (Alsheikh et al 2022). TDM queries were focused on paragraphs with clinical / human focus.…”

Glucocorticoid-regulated bidirectional enhancer RNA transcription pinpoints functional genetic variants linked to asthma

“…However, the sum of heritabilities explained by these SNPs is estimated to account for only 2.5% of the estimated genetic risk of asthma [5][6][7] . Moreover, the majority of asthma-associated SNPs defined though GWAS are in non-protein coding portions of the genome 8 , and direct regulatory effects of most such SNPs have not been established 9 . These relative weaknesses of GWAS-based approaches are likely due to several underlying factors, including unaccounted effects of environment on asthma risk, linkage disequilibrium, and the very stringent p-values needed to identify significant genetic associations on a genomewide basis 10 .…”