Genetic, epigenomic, and functional approaches to identifying effector genes for complex diseases

You are invited to join us on Monday, October 18 at 1pm US Eastern time for an ASHG 2021 ancillary session presented by the Common Metabolic Diseases Knowledge Portal. Kick off your ASHG week with a dynamic session featuring short talks from Anna Gloyn, Brent Richards, Krishna Aragam, and Jason Flannick on methods for identifying and characterizing effector genes for common metabolic diseases.

To receive a calendar invitation with connection information, please register for the webinar. Or, connect using the links below.

Genetic, epigenomic, and functional approaches to identifying effector genes for complex diseases.
Monday, October 18, 2021 1:00 - 2:30 PM Eastern Time (US and Canada)

Connection information:

Join the webinar
Passcode: 971794
Or One tap mobile : 
    US: +16468769923,,87230394607#,,,,*971794#  or +16465588656,,87230394607#,,,,*971794# 
Or Telephone:
    Dial (for higher quality, dial a number based on your current location):
        US: +1 646 876 9923  or +1 646 558 8656 
Webinar ID: 872 3039 4607
Passcode: 971794
International numbers available



Noël Burtt
Director, Operations and Development, Diabetes Research and Knowledge Portals
Broad Institute of MIT and Harvard

Translational functional genomics of diabetes as a model for complex disease research 

Anna Gloyn
Professor of Pediatrics (Endocrinology) and, by courtesy, of Genetics
Stanford University

Genome-wide association studies (GWAS) have identified thousands of signals robustly associated with risk for common diseases. At the vast majority of these loci, the lead single nucleotide polymorphisms (SNPs) reside in noncoding regions of the genome, which hampers biological inference and translation of genetic discoveries into disease mechanisms. Using type 2 diabetes as a model, this talk will describe these challenges and illustrate how a holistic approach involving genetics, epigenomics, physiology, and cellular and developmental biology can identify disease effector genes and uncover disease mechanisms.

An Effector Index to Identify Causal Genes at GWAS Loci

Brent Richards
Professor, Departments of Medicine, Human Genetics, Epidemiology and Biostatistics
McGill University

Drug development and biological discovery require effective strategies to map existing genetic associations to causal genes. To approach this problem, we identified a set of positive control genes for 12 common diseases and traits that cause a Mendelian form of the disease or are the target of a medicine used for disease treatment. We then identified a simple set of genomic features enriching GWAS-associated single nucleotide variants (SNVs) for these positive control genes. Using these features, we trained and validated the Effector Index (Ei), a causal gene mapping algorithm using the 12 common diseases and traits. The area under Ei’s receiver operator curve to identify positive control genes was 80% and area under the precision recall curve was 29%. Using an enlarged set of independently curated positive control genes for type 2 diabetes which included genes identified by large-scale exome sequencing, these areas increased to 85% and 61%, respectively. The best predictors were coding or transcript altering SNVs, distance to gene and open chromatin-based metrics. We have developed the Ei algorithm for its widespread use. Ei provides a simple, understandable tool to prioritize genes at GWAS loci for functional follow-up and drug development. Results can be explored via the Accelerating Medicines Partnership Common Metabolic Diseases Knowledge Portal Predicted Effector Genes page.

Integrated and curated approaches to generating effector lists for CAD

Krishna Aragam
Instructor in Medicine
Massachusetts General Hospital

To address critical barriers to the identification of causal genes for coronary artery disease (CAD), we applied disease-specific, complementary “locus-based” and “similarity-based” strategies for gene prioritization, to systematically identify likely effector genes across 241 genome-wide significant associations with CAD. Specifically, we integrated eight objective locus-based or similarity-based features predictive of causal genes: (1) genomic distance; (2) monogenic disorders of cardiovascular relevance; (3) rare coding variants previously associated with CAD risk; (4) protein-altering variants (missense or putative loss-of-function); (5) proteins of causal relevance to CAD per Mendelian randomization, or targets of established cardiovascular drugs; (6) eQTLs in CAD-relevant tissues from STARNET or GTEx; (7) prioritization by a new similarity-based method, the polygenic priority score (PoPS); and (8) cardiovascular-relevant phenotypes when knocked-out in mouse models from the International Mouse Phenotyping Consortium or Mouse Genome Informatics database. 206 associations (85.5%) had a gene that was prioritized by two or more features, with 47 associations having at least four features pointing to the same causal gene. Strongly prioritized causal genes included those well-established in CAD pathogenesis (e.g. LDLR, LPL, PCSK9, NOS3, ANGPTL4, APOE, GUCY1A3), confirming the validity of our approach. For 32 (16.5%) of the associations the nearest gene to the sentinel variant was not prioritized, including associations at APOC3, PLTP, and TGFB1. Our analysis provides a catalogue of candidate genes prioritized by orthogonal and disease-specific lines of evidence to inform experimental interrogation of putative causal mechanisms for CAD, and will be maintained as a community resource through the AMP Cardiovascular Disease Knowledge Portal (

Providing access to effector gene predictions for complex disease researchers

Jason Flannick
Assistant Professor of Pediatrics, Harvard Medical School and the Division of Genetics and Genomics at Boston Children’s Hospital
Associate member, Broad Institute of MIT and Harvard

Approaches for predicting GWAS effector genes have applications in many research domains. They are therefore useful to researchers of many backgrounds and from many communities. Current gaps in the accessibility of effector gene predictions include: (a) predictions are unavailable for many diseases or, if available, are often known only internally within a small community; (b) different prediction approaches can disagree, making it challenging to determine which are correct; and (c) statistical methods for summarizing predictions across multiple approaches are in their infancy. In this seminar, we will present a new resource for accessing effector gene predictions for many complex traits. We will describe methods we have developed for scaling both bioinformatic and curation approaches to hundreds of different diseases. We will finally present new statistical approaches for combining different prediction algorithms into summaries of evidence for each potential effector gene, as well as a web-based visualization for understanding the data underlying each prediction. Attendees will learn how to access effector gene predictions for a broad variety of traits of interest, and they will also learn tools and guidelines to evaluate how confident they can be in an effector gene prediction.