RT Journal Article
SR Electronic
T1 Learning causal biological networks with generalized Mendelian randomization
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 171348
DO 10.1101/171348
A1 Md. Bahadur Badsha
A1 Audrey Qiuyan Fu
YR 2017
UL http://biorxiv.org/content/early/2017/08/01/171348.abstract
AB Learning causality from biological data remains a challenge. We present MRPC, a novel machine learning algorithm that employs generalized Mendelian randomization and learns a causal biological network with directed edges. Our method has several desirable statistical features: it controls the false discovery rate, and performs robust inference. Using MRPC, we distinguished direct and indirect targets among multiple genes associated with eQTLs, and constructed a network for frequently altered cancer genes.