An efficient algorithm for identifying mutated subnetworks associated with survival in cancer.

Authors:
Ivan Arisi
Ivan Arisi
Institute of Neurobiology and Molecular Medicine
Italy
Cesare Saltini
Cesare Saltini
University of Rome Tor Vergata
Italy
Tamer Kahveci
Tamer Kahveci
University of Florida
United States

IEEE/ACM Trans Comput Biol Bioinform 2019 Apr 15. Epub 2019 Apr 15.

Protein-protein interaction (PPI) network models interconnections between protein-encoding genes. Group of proteins that perform similar functions are often connected to each other in PPI network. The corresponding genes form pathways or functional modules. Mutation in protein-encoding genes affect behavior of pathways. This results in initiation, progression and severity of diseases that propagates through pathways. In this work, we integrate mutation, survival information of patients and PPI network to identify connected subnetworks associated with survival. We define computational problem using fitness function called log-rank statistic to score subnetworks. Log-rank statistic compares the survival between two populations. We propose a novel method, Survival Associated Mutated Subnetwork (SAMS) that adopts genetic algorithm strategy to find connected subnetwork within PPI network whose mutation yields highest log-rank statistic. We test on real cancer and synthetic datasets. SAMS generate solutions in negligible time while state-of-art method in literature takes exponential time. Log-rank statistic of SAMS selected mutated subnetworks are comparable to the method. Our result genesets show significant overlap with well-known cancer driver genes derived from curated datasets and studies in literature, display high text-mining score in terms of number of citations combined with disease-specific keywords in PubMed and identify pathways having high biological relevance.

Download full-text PDF

Source
http://dx.doi.org/10.1109/TCBB.2019.2911069DOI Listing
April 2019
1 Read

Publication Analysis

Top Keywords

ppi network
16
log-rank statistic
16
associated survival
8
subnetworks associated
8
mutated subnetworks
8
protein-encoding genes
8
survival
5
connected subnetwork
4
subnetwork ppi
4
network mutation
4
genetic algorithm
4
find connected
4
algorithm strategy
4
strategy find
4
mutation yields
4
test real
4
synthetic datasets
4
datasets sams
4
sams generate
4
generate solutions
4

Similar Publications