TPdsm: a method based on TabPFN for prediction of deleterious synonymous mutations

Area of Science:

• Genomics
• Computational Biology
• Human Health

Background:

• Synonymous mutations are crucial for understanding human health.
• Predicting deleterious synonymous mutations is vital but hindered by limited training data.

Purpose of the Study:

• To introduce TPdsm, a novel method for predicting deleterious synonymous mutations.
• To address the challenge of data scarcity in developing accurate prediction models.

Main Methods:

• TPdsm utilizes TabPFN, a tabular foundation model adept at small-sample predictions.
• Features are sourced from CDsyn, a specialized database for deleterious synonymous mutations.

Main Results:

• TPdsm demonstrates superior predictive performance compared to 14 state-of-the-art predictors.
• Validation across multiple independent datasets and real-world cases confirms TPdsm's efficacy.

Conclusions:

• TPdsm offers a high-performance solution for predicting deleterious synonymous mutations.
• The method effectively overcomes the bottleneck of limited training data, advancing genomic analysis for health insights.