Predicting gene structure changes resulting from genetic variants via exon definition features.

Area of Science:

• Genomics
• Computational Biology
• Molecular Genetics

Background:

• Genetic variations can alter gene splicing, impacting traits and diseases.
• Existing gene prediction tools struggle with individual genetic variations affecting splicing.
• Reference gene finders assume conserved, functional gene structures, limiting their accuracy for individual variations.

Purpose of the Study:

• To develop a probabilistic computational approach for predicting gene structure changes, including those that alter splicing.
• To assess the feasibility of gene structure prediction without relying on conserved coding features.
• To investigate the impact of genetic variations on splicing patterns in individual human genomes.

Main Methods:

• A probabilistic model was developed to predict gene structure alterations, applicable to both coding and non-coding genes.
• The model was trained on existing gene annotations without requiring curated aberrant splicing examples.
• Gene structure prediction was performed on individual human genomes, specifically focusing on altered splicing patterns.

Main Results:

• The model successfully predicted gene structure changes without relying on conserved coding features.
• An unexpected abundance of variants creating de novo splice sites was identified.
• These de novo splice variants, often overlooked, can significantly affect splicing and protein products, potentially acting as cryptic disease factors.

Conclusions:

• Predicting gene structure changes, particularly those affecting splicing, is feasible even without assuming conserved coding features.
• De novo splice variants represent a significant class of genetic variation with potential roles in disease.
• The developed method offers a valuable tool for understanding the functional impact of genetic variation on gene splicing.