Predicting exon criticality from protein sequence

Area of Science:

• Genomics
• Proteomics
• Bioinformatics

Background:

• Alternative splicing diversifies protein function and presents therapeutic opportunities.
• Predicting exon skipping is crucial for understanding gene regulation and disease.
• Current methods often rely on intronic sequences or tissue-specific data.

Purpose of the Study:

• To develop a predictive model, Exon ByPASS, for assessing exon inclusion criticality using only protein amino acid sequences.
• To predict exon skipping independent of tissue and species, without intronic information.
• To identify potential therapeutic targets by predicting skippable exons and neo-junctions in cancer.

Main Methods:

• Developed Exon ByPASS, a machine learning model utilizing amino acid sequences flanking exon junctions.
• Focused on protein sequence information to predict exon skipping events.
• Validated model predictions against transcriptomic and proteomic datasets.

Main Results:

• Exon ByPASS accurately predicts exon skipping based solely on protein sequence information.
• The model demonstrates cross-species and cross-tissue applicability.
• Identified synthetically skippable exons and neo-junctions in cancer, suggesting therapeutic potential.

Conclusions:

• Exon ByPASS offers a novel, sequence-based approach to predict exon skipping.
• The model has broad applications in basic research and therapeutic development.
• Predicting exon skipping via protein sequence analysis opens new avenues for cancer therapy.