Premium
A modified PageRank algorithm for biological pathway ranking
Author(s) -
Zhang Qingyang
Publication year - 2018
Publication title -
stat
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 18
ISSN - 2049-1573
DOI - 10.1002/sta4.204
Subject(s) - ranking (information retrieval) , computational biology , pagerank , biological pathway , computer science , phenotype , bioinformatics , artificial intelligence , biology , gene , theoretical computer science , genetics , gene expression
Pathways are the functional building blocks of complex diseases such as cancer. Identifying disease‐associated pathways is of great importance to the development of novel therapeutics, as it provides functional insights into the pathogenesis of a disease. Existing methods for pathway ranking, however, are mostly based on an enrichment score assigned to each pathway independently, which could be biased by overlooking the interactions between pathways. In this paper, we consider a modification of Google's PageRank algorithm in order to fully incorporate the pathway dependencies into the pathway ranking. The proposed measurement is a trade‐off between two important aspects, namely, the phenotype–pathway association and pathway coexpression. We propose to use a projection correlation to quantify pathway coexpression, and a generalized R 2 for phenotype–pathway association. Our simulation study on real pathways shows the competitive performance of the new measure compared with enrichment‐based analyses in pathway ranking.