Rule-based network optimization to infer dysregulated signaling from -omics data

点击翻译按钮获取中文摘要

Pathway analyses of omic data provide mechanistic insights which facilitate interpretation. Current pathway analysis approaches, however, are unable to distinguish between pathways which have divergent signal origin but common effector molecules because solutions are exclusively based on static properties. Sequential dynamical systems (SDS) modeling allows inference of dynamics in pathway analysis. Further, by capturing emergent phenomena in molecular networks, dynamic approaches to drug re-purposing facilitate in silico experimentation and investigation of non-target effects. A key hindrance to use of SDS models with omic data has been modeling variance within omic data as arising from intracellular stochasticity rather than cellular heterogeneity. To this end, I will develop methodology that accounts for heterogenous cell states in bulk omics data, and re-implement extant inference techniques to recover necessary and sufficient conditions for underlying network transitions. This will be accomplished by implementing Boolean update models, which take molecules as either active or inactive, across an ensemble of starting states to construct Ensemble Boolean Networks (EBN). EBNs will improve dynamic simulations of molecular networks and in-silico perturbation analysis. Specifically, EBN algorithms will then be applied in parallel with existing SDS algorithms to perform network-based pathway analysis of omics data and to investigate dysregulated signaling subnetworks in disease states for drug re-purposing. An SDS-based pathway-level metric that explicitly considers interactions between molecules will be achieved by perturbation analysis of pathway components followed by development of a pathway-level score based on a weighted node-level metric. I will use this technique to help our collaborators gain insight into placental biology and B cell migration using transcriptomic and proteomic datasets, respectively. An SDS- based algorithm to repurpose FDA approved drugs using omic data from drug-treated and disease-perturbed states will be assembled by quantifying signaling dysregulation in disease states from transcriptomic data in public domain. This technique will be applied to understand dysregulation of platelets and monocytes in the development of atherosclerosis in people living with HIV. SDS-based pathway analysis will improve the prediction of key nodes in pathways, facilitating translation of omic data into in vivo and in vitro studies. SDS- based repurposing will provide a powerful new way to combine prior knowledge, extant drug omic data, and extant disease omic data to uncover new potential therapeutic agents. Taken together, this proposal will develop a new technique called EBN and will apply it alongside other SDS-based techniques to generate innovative algorithms to retrieve key features and their regulatory context from omic datasets.