ABI INNOVATION: Molecular Network Inference via Nonparametric Functional Dependencies

New Mexico State University, Las Cruces NM

Investigators

Abstract

Despite sharing a common genome, cells and tissues making up the same organism can have distinct biological function. One explanation is that the molecular networks in which genes interact are wired differently across cell types within the organism. The purpose of this project is to identify the functional molecular networks in major cell and tissue types in two mammals, humans and mice. Using gene expression, or transcriptome, data with newly developed statistical inference methodologies, this research will uncover the connections currently obscured by the complex gene interaction patterns that result from the concurrent activity of many networks. There are rich 'omics data resources profiling these dynamic molecular rewiring processes, including the FANTOM5 promoter-level gene expression atlases of mammalian primary cells and tissues, and tissue-specific human and mouse epigenome profiles from the ENCODE project. Using these public data sets and innovative computational methods, the project is expected to decipher epigenetically rewired molecular networks across a number of types of tissues during differentiation. The resulting high-resolution networks can then be compared between the organisms, showing how functional networks have evolved between human and mouse. Therefore, the project is expected to be transformative in the understanding of complex biological mechanisms, implementing a strategy demonstrating the causality-by-functionality principle. Network inference is becoming a bottleneck in modern biology as high-throughput data backlogs have surpassed the capacity of human interpretation. To address this challenge, the goal of this project is to design a nonparametric functional dependency framework for network inference using the causality-by-functionality principle, apply it in characterizing mammalian molecular networks, and evaluate the performance in a subnetwork underlying cerebellar development. With the latest excellent mammalian transcriptome and epigenome resources, the project will reveal combinatorial functional effects to understand causal molecular interactions, detect tissue-specific network rewiring in mammalian cell differentiation, characterize pathway evolution between human and mouse at the promoter level, and validate a three-layer inferred gene subnetwork for cerebellar development. The project outcomes are expected to include molecular networks involving transcription factors, chromatin remodelers, and noncoding RNAs; second-order network rewiring across major cell and tissue types in mammalian cell differentiation; and pathway evolution between human and mouse in response to an ongoing debate on transcriptome divergence between the two species. The inferred cerebellar subnetwork after validation will contribute to new biology in brain development research. The network inference methodologies are expected to substantially expand the landscape of discoverable complex nonlinear interactions in dynamic networks, and the resulting software will be an essential informatics tool for discoveries in biology. Educational activities planned at New Mexico State University will enhance biological informatics workforce development and prepare students for knowledge-based career opportunities. Project information is available at https://www.cs.nmsu.edu/~joemsong/funchisq .