MTSplice

MTSplice predicts the tissue-specific effects of genetic variants on splicing of cassette exons across 56 human tissues.

Key Features:

• Neural network model: A neural network models tissue-specific regulatory sequences to predict variant effects on cassette exon splicing.
• Integration with MMSplice: Incorporates MMSplice predictions for constitutive regulatory sequences alongside tissue-specific modeling.
• Quantitative tissue profiling: Produces quantitative predictions of variant-induced changes in cassette exon splicing across 56 tissues.
• Validation on GTEx: Tested and benchmarked using naturally occurring genetic variants from the GTEx dataset, showing improved performance over MMSplice across most tissues.
• Disease-focused analyses: Applied to assess enrichment of autism-associated de novo mutations affecting splicing specifically in brain tissues.

Scientific Applications:

• Functional genomics: Characterizing how genetic variation alters tissue-specific splicing and gene regulation.
• Disease research: Identifying tissue-specific splicing changes that may contribute to disorders such as neurological conditions including autism.
• Variant prioritization: Prioritizing genetic variants for experimental follow-up based on predicted impact on cassette exon splicing in specific tissues.

Methodology:

Developed as a neural network model for tissue-specific regulatory sequences integrated with MMSplice and tested on naturally occurring genetic variants from the GTEx dataset to predict effects on cassette exon splicing across 56 tissues.