UniSplicer: A deep-learning framework for accurate splice-site prediction and splice-altering mutation detection...

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Accurate splice site identification is crucial for understanding gene structure and function.
• Existing gene annotation methods require extensive resources and high-quality genome assemblies, limiting their application in non-model species.
• Limited transcriptomic data poses a challenge for splice site prediction in many organisms.

Purpose of the Study:

• To develop a deep learning-based framework for accurate intron splice site prediction across diverse species, even with limited transcriptomic data.
• To create a tool that overcomes resource limitations associated with traditional gene annotation methods.
• To provide a reliable method for analyzing splicing alterations in large genomic datasets.

Main Methods:

• Development of a deep learning-based model training framework named UniSplicer.
• Training and validation of UniSplicer models on transcriptomic data from various species (plants, fungi, metazoans).
• Comparative analysis of UniSplicer performance against existing splice site prediction models.

Main Results:

• UniSplicer-based models demonstrated superior prediction accuracy compared to existing methods across diverse species.
• Prediction scores from UniSplicer models serve as reliable indicators for the impact of mutations on splicing.
• The UniSplicer Arabidopsis thaliana model identified genes with splicing abnormalities due to sequence variations, suggesting potential environmental selection.

Conclusions:

• UniSplicer provides a highly accurate and resource-efficient solution for intron splice site prediction in a wide range of species.
• The framework facilitates the analysis of splicing alterations and mutation effects in large genomic datasets, particularly for non-model organisms.
• UniSplicer offers valuable insights into the relationship between sequence variations, splicing, and potential environmental adaptation.