NetMix2: Unifying Network Propagation and Altered Subnetworks

Uthsav Chitra, Tae Yoon Park, Benjamin J. Raphael

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations


A standard paradigm in computational biology is to use interaction networks to analyze high-throughput biological data. Two common approaches for leveraging interaction networks are: (1) network ranking, where one ranks vertices in the network according to both vertex scores and network topology; (2) altered subnetwork identification, where one identifies one or more subnetworks in an interaction network using both vertex scores and network topology. The dominant approach in network ranking is network propagation which smooths vertex scores over the network using a random walk or diffusion process, thus utilizing the global structure of the network. For altered subnetwork identification, existing algorithms either restrict solutions to subnetworks in subnetwork families with simple topological constraints, such as connected subnetworks, or utilize ad hoc heuristics that lack a rigorous statistical foundation. In this work, we unify the network propagation and altered subnetwork approaches. We derive a subnetwork family which we call the propagation family that approximates the subnetworks ranked highly by network propagation. We introduce NetMix2, a principled algorithm for identifying altered subnetworks from a wide range of subnetwork families, including the propagation family, thus combining the advantages of the network propagation and altered subnetwork approaches. We show that NetMix2 outperforms network propagation on data simulated using the propagation family. Furthermore, NetMix2 outperforms other methods at recovering known disease genes in pan-cancer somatic mutation data and in genome-wide association data from multiple human diseases. NetMix2 is publicly available at

Original languageEnglish (US)
Title of host publicationResearch in Computational Molecular Biology - 26th Annual International Conference, RECOMB 2022, Proceedings
EditorsItsik Pe’er
PublisherSpringer Science and Business Media Deutschland GmbH
Number of pages16
ISBN (Print)9783031047480
StatePublished - 2022
Event26th International Conference on Research in Computational Molecular Biology, RECOMB 2022 - San Diego, United States
Duration: May 22 2022May 25 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13278 LNBI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference26th International Conference on Research in Computational Molecular Biology, RECOMB 2022
Country/TerritoryUnited States
CitySan Diego

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science


  • Cancer
  • GWAS
  • Interaction networks
  • Network anomaly
  • Network propagation


Dive into the research topics of 'NetMix2: Unifying Network Propagation and Altered Subnetworks'. Together they form a unique fingerprint.

Cite this