Year: 2015
Type: article
Abstract: Biomarkers are the actionable factors differentiating any condition states, specifically disease response and non-response to treatment.They have been used as screening measures to indicate surrogate or subclinical manifestations, and are also prime targets of drug repositioning approaches.We aim to identify gene sub-networks involved in response to a drug by implementing a data-mining algorithm that identifies genes with highly correlated expression across the population, concomitantly significantly variable between response and non-response cohorts to allow differentiation.Our greedy Quasi-Clique Merger implementation, localMax-eQCM, is tuned for identifying unique subnetwork modules by restricting initial gene pairs to only those of highest local coexpression (a rare event), thereafter keeping with the original QCM, that each and every gene within the sub-network be co-expressed to a high degree (Figure 2.3).We are then able to differentiate differentially expressed modules between response subpopulations by using a t-test among respective eigengene.This method of biomarker evaluation could ultimately be employed to assign patients to receive chemotherapy, predicted to productively respond based upon biomarker screening.We implement our algorithm across CCLE gene expression from a NSCLC large and squamous cell subpopulation, of which cell lines were immortalized and subsequent Taxol response profiles (IC 50 ) measured.We identified three sub-network modules highly correlated to Taxol response in this way.iii Dedication For Bud
Author: Daniel Morgan
Institution: The Ohio State University
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2735225934