TY - GEN
T1 - Multitask matrix completion for learning protein interactions across diseases
AU - Kshirsagar, Meghana
AU - Carbonell, Jaime G.
AU - Klein-Seetharaman, Judith
AU - Murugesan, Keerthiram
N1 - Publisher Copyright:
© Springer International Publishing Switzerland 2016.
PY - 2016
Y1 - 2016
N2 - Disease causing pathogens such as viruses, introduce their proteins into the host cells where they interact with the host’s proteins enabling the virus to replicate inside the host. These interactions between pathogen and host proteins are key to understanding infectious diseases. Often multiple diseases involve phylogenetically related or biologically similar pathogens. Here we present a multitask learning method to jointly model interactions between human proteins and three different, but related viruses: Hepatitis C, Ebola virus and Influenza A. Our multitask matrix completion based model uses a shared low-rank structure in addition to a task-specific sparse structure to incorporate the various interactions. We obtain upto a 39% improvement in predictive performance over prior state-of-the-art models.We show how our model’s parameters can be interpreted to reveal both general and specific interactionrelevant characteristics of the viruses. Our code, data and supplement is available at: http://www.cs.cmu.edu/∼mkshirsa/bsl mtl.
AB - Disease causing pathogens such as viruses, introduce their proteins into the host cells where they interact with the host’s proteins enabling the virus to replicate inside the host. These interactions between pathogen and host proteins are key to understanding infectious diseases. Often multiple diseases involve phylogenetically related or biologically similar pathogens. Here we present a multitask learning method to jointly model interactions between human proteins and three different, but related viruses: Hepatitis C, Ebola virus and Influenza A. Our multitask matrix completion based model uses a shared low-rank structure in addition to a task-specific sparse structure to incorporate the various interactions. We obtain upto a 39% improvement in predictive performance over prior state-of-the-art models.We show how our model’s parameters can be interpreted to reveal both general and specific interactionrelevant characteristics of the viruses. Our code, data and supplement is available at: http://www.cs.cmu.edu/∼mkshirsa/bsl mtl.
KW - Host-pathogen protein interactions
KW - Matrix completion
KW - Multitask learning
UR - http://www.scopus.com/inward/record.url?scp=84964078244&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84964078244&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-31957-5_4
DO - 10.1007/978-3-319-31957-5_4
M3 - Conference contribution
AN - SCOPUS:84964078244
SN - 9783319319568
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 53
EP - 64
BT - Research in Computational Molecular Biology - 20th Annual Conference, RECOMB 2016, Proceedings
A2 - Singh, Mona
PB - Springer Verlag
T2 - 20th Annual Conference on Research in Computational Molecular Biology, RECOMB 2016
Y2 - 17 April 2016 through 21 April 2016
ER -