PCLassoLog: A protein complex-based, group Lasso-logistic model for cancer classification and risk protein complex discovery.

ABSTRACT: Risk gene identification has attracted much attention in the past two decades. Since most genes need to be translated into proteins and cooperate with other proteins to form protein complexes to carry out cellular functions, which significantly extends the functional diversity of individual proteins, revealing the molecular mechanism of cancer from a comprehensive perspective needs to shift from identifying individual risk genes toward identifying risk protein complexes. Here, we embed protein complexes into the regularized learning framework and propose a protein complex-based, group Lasso-logistic model (PCLassoLog) to discover risk protein complexes. Experiments on deep proteomic data of two cancer types show that PCLassoLog yields superior predictive performance on independent datasets

SUBMITTER: Wang W 

PROVIDER: S-EPMC9791601 | biostudies-literature | 2023

REPOSITORIES: biostudies-literature