A shiny app that integrates protein-protein interactions, clinical variants, & protein structure to enhance the interactome.
Most existing PPI databases only curate a binary interactome. However, integrating protein structure is critical in understanding complex human diseases because variants (GOF/LOF) have significant impacts on protein structure. Thus, having a more complete interactome will aid in the identification of pathogenic variants.
SNVariome is a tool that seeks to integrate protein structure, protein-protein interactions, drug-gene interaction data, variant data (using ClinVar, GOF/LOF data), and phenotype data (HPO terms) to develop a comprehensive interactome of clinical variants. We will utilize the SigCom LINCS API and integrate a publicly available dataset that has already examined GOF/LOF impacts on protein structure (pmcid: PMC9259657). Our first test case will be a list of genes/variants associated with breast cancer which should validate and expand the findings of DOI: 10.1126/science.abf3066.
During Hackin' Omics, we'd like to:
Run our shiny app either using the hosted version or via the rconsole.
The input is a gene or variant and the output is signature data, edc values, and PPIs.
install.packages(c('shiny', 'BiocManager', 'shinyjs', 'bslib', 'httr', 'jsonlite', 'xml2', 'DT', 'reticulate'))
BiocManager::install(c('wppi', 'myvariant'))
To run this app, open RStudio and use the below code in the console:
library(shiny)
shiny::runGitHub(username = "u-brite", repo = "SNVariome", ref = "main", subdir = "app")
After using the human protein interactome to find protein-protein interactions across a list of clinic variants classified as either gain of function, loss of function, or haploinsufficient, we found that out of the over 1200 genes indicated in protein interaction, only 104 of the 135 with autosomal dominant inheritance were found to have interactions in the interactome. We chose autosomal dominant genes since there is a highly significant tendency for disease mutations in AD genes to be more clustered.
While applying our methods, we found that we were able to see that our networks supported the literature and ultimately extended the knowledge-base by adding data regarding signatures. Specifically, looking at BRCA1 (breast cancer) has shown us novel signatures and interactions not seen in literature up until our analysis.
Shaurita Hutchins | Team leader
Bernhard Hane | Team member
Oladosu Tosin Ayodeji | Team member
Bharat Mishra | Team member
Hailey Levi | Team member
Pooja Singaravelu | Team member
Maria Jose | Team member
Gerasimavicius L, Livesey BJ, Marsh JA. Loss-of-function, gain-of-function and dominant-negative mutations have profoundly different effects on protein structure. Nat Commun. 2022 Jul 6;13(1):3895. doi: 10.1038/s41467-022-31686-6. PMID: 35794153; PMCID: PMC9259657.
Kim M, Park J, Bouhaddou M, Kim K, Rojc A, Modak M, Soucheray M, McGregor MJ, O'Leary P, Wolf D, Stevenson E, Foo TK, Mitchell D, Herrington KA, Muñoz DP, Tutuncuoglu B, Chen KH, Zheng F, Kreisberg JF, Diolaiti ME, Gordan JD, Coppé JP, Swaney DL, Xia B, van 't Veer L, Ashworth A, Ideker T, Krogan NJ. A protein interaction landscape of breast cancer. Science. 2021 Oct;374(6563):eabf3066. doi: 10.1126/science.abf3066. Epub 2021 Oct 1. PMID: 34591612; PMCID: PMC9040556.