TraitProtNet: Deciphering the Genome for Trait Prediction with Interpretable Deep Learning

Abstract

Genome data is far from fully explored. We present TraitProtNet, an innovative deep learning framework for predictive trait profiling in fungi, leveraging genome data and pretrained language models. The use of Integrated Gradients and bioinformatic analysis provides insights into the interpretability of the model, complementing traditional omics by highlighting the difference between protein importance and expression levels. This framework offers significant potential for future applications in both agriculture and medicine.