Multiobjective Genome-Wide RNA-Binding Event Identification From CLIP-Seq Data.

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• RNA-binding proteins (RBPs) regulate gene expression post-transcriptionally.
• Crosslinking immunoprecipitation sequencing (CLIP-seq) generates genome-wide RNA-binding data.
• Existing methods for analyzing CLIP-seq data face challenges like high costs and computational intensity.

Purpose of the Study:

• To develop an efficient computational method for identifying protein-RNA interactions from CLIP-seq data.
• To address limitations of current methods, including cost, computation, dimensionality, instability, and sparsity.
• To improve the understanding of RBP functions in cellular processes.

Main Methods:

• A novel computational method, the multiobjective forest algorithm (MFA), was developed.
• MFA integrates multiobjective biogeography-based optimization (BBO) with random forest (RF).
• Multiobjective BBO dynamically prunes RF classifiers, balancing model generality and complexity.

Main Results:

• MFA demonstrated superior performance compared to state-of-the-art methods across 31 CLIP-seq datasets.
• The method provides mechanistic insights through data source importance analysis and other evaluations.
• Experimental results validate MFA's robustness and effectiveness in identifying protein-RNA interactions.

Conclusions:

• The multiobjective forest algorithm (MFA) offers a powerful and efficient approach for analyzing CLIP-seq data.
• MFA overcomes limitations of existing methods, enabling more accurate identification of RNA-binding events.
• This advancement facilitates a deeper understanding of RNA-binding protein functions and gene regulation.