Host genes that help coronaviruses, deciphering missense variants, and machine learning delves into bacteria's sweet side
Research Roundup: October 30, 2020
Welcome to the October 30, 2020 installment of Research Roundup, a recurring snapshot of recent studies published by scientists at the Broad Institute and their collaborators.
COVID collaborators among our genes
Key to combating COVID-19 is a clearer picture of how our own genes contribute to the disease. Peter DeWeirdt, Ruth Hanna, institute scientist John Doench, and others in the Genetic Perturbation Platform teamed up with with Craig Willen (Yale) and colleagues to run genome-wide CRISPR screens against the viruses behind COVID-19, MERS, and the 2003 SARS outbreak, looking for host genes that influence coronavirus infection. They report in Cell that HMGB1 and SWI/SNF protein complex members are pro-viral, while the histone H3 complex plays an anti-viral role. Learn more in a Yale/Broad news story.
Making sense of missense variants
Many missense variants involve amino acid substitutions, and interpreting their effects on protein structure and function is challenging. Using more than 14,000 experimentally solved human protein structures, research scientist Sumaiya Iqbal and group leader Arthur Campbell of the Center for the Development of Therapeutics, Dennis Lal of the Stanley Center for Psychiatric Genetics, and colleagues have characterized the amino acid positions affected in missense variants from 1,330 disease-associated genes. In a study in PNAS, the team describes key features associated with 33,000 pathogenic missense variants, 165,000 variants found in the general population, and unique characteristics of altered amino acid positions in 24 protein classes. They also show that variants' function-specific 3D features match the readouts of mutagenesis experiments for BRCA1 and PTEN. Explore all the data here.
Host-microbe interactions can put evolutionary pressure on glycans, the most diverse biopolymer. In Cell Host & Microbe, Daniel Bojar (Wyss/MIT), Rani Powers (Wyss/MIT), Diogo Camacho (Wyss), and institute member Jim Collins of MIT, the Wyss Institute, and the Broad’s Infectious Disease and Microbiome Program, present a resource toolkit comprising machine learning and bioinformatics methods with a large glycan database to leverage evolutionary information present in glycans. Based on a deep-learning language model called SweetTalk, the toolkit can enable glycan-based studies of host-microbe interactions.