CeModule: an integrative framework for discovering regulatory patterns from genomic data in cancer.

Area of Science:

• * Computational biology
• * Molecular oncology
• * Systems biology

Background:

• * Non-coding RNAs (ncRNAs) are critical regulators in tumorigenesis.
• * Long non-coding RNAs (lncRNAs) can interact with microRNAs (miRNAs) via competing endogenous RNA (ceRNA) mechanisms.
• * Understanding lncRNA-miRNA-mRNA interactions is crucial for deciphering disease pathogenesis and identifying therapeutic targets.

Purpose of the Study:

• * To develop an integrative computational framework, CeModule, for discovering regulatory modules involving lncRNAs, miRNAs, and mRNAs.
• * To model competing endogenous RNA (ceRNA) interactions at the post-transcriptional level.
• * To enhance understanding of complex regulatory networks in human diseases.

Main Methods:

• * Developed CeModule, an integrative framework utilizing matched expression profiles of lncRNAs, miRNAs, and mRNAs.
• * Employed joint orthogonality non-negative matrix factorization for module identification.
• * Incorporated experimentally verified miRNA-lncRNA and miRNA-mRNA interactions, along with gene-gene networks, using network-based penalties and sparse regularizations.
• * Utilized an iterative multiplicative updating algorithm to solve the optimization problem.

Main Results:

• * Identified significant regulatory modules comprising lncRNAs, miRNAs, and mRNAs.
• * Modules were found to be significantly associated with cancer-related biological processes and pathways.
• * Demonstrated the framework's ability to uncover complex regulatory patterns.

Conclusions:

• * CeModule successfully identified functionally enriched modules in ovarian cancer (OV) and uterine corpus endometrial carcinoma (UCEC) datasets.
• * The findings provide novel insights into the systems-level regulatory mechanisms of human diseases.
• * This approach may facilitate the discovery of new diagnostic and therapeutic strategies for cancer.