RT Journal Article
SR Electronic
T1 Network inference from single-cell data using multivariate information measures
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 082099
DO 10.1101/082099
A1 Thalia E. Chan
A1 Michael P.H. Stumpf
A1 Ann C. Babtie
YR 2017
UL http://biorxiv.org/content/early/2017/04/26/082099.abstract
AB Gene expression is robustly controlled in space and time by networks of transcriptional regulators. Single cell data now allow us to probe and map these networks and cellular transcriptional states in unprecedented detail, but the heterogeneity and large sample sizes that characterise these data pose new analytical challenges. Here we develop an algorithm that exploits these features to infer gene (co-)regulatory networks from single cell data using multivariate information measures. Information theory allows us to detect complex non-linear relationships and explore statistical dependencies between multiple genes in detail. We thoroughly evaluate the performance of our algorithm in comparison to existing methods, demonstrate the benefits of considering multivariate information measures, and illustrate its application to several datasets. We provide guidelines for the use of information theoretical based algorithms, and software implementing our method to enable the identification of putative functional relationships and mechanistic hypotheses from single cell transcriptomic data.