A Novel Method For Identifying Key Genes In Macroevolution Based On Deep Learning With Attention Mechanism

This paper presents IKGM, a method based on deep learning with attention mechanisms for identifying key genes in the macroevolution of biological taxa. Using 34 species of diurnal butterflies and nocturnal moths as an example, IKGM mines the key genes with high weights and performs KEGG enrichment analysis based on these genes. The pipeline of this paper consists of four parts: Data pre-processing, Classification Model, Weights calculation, and Evaluation. Three types of attention mechanisms (domain attention, kmer attention and fused attention) were developed to calculate the weights of different genes.