A novel method for identifying key genes in macroevolution based on deep learning with attention mechanism

Area of Science:

• Evolutionary Biology
• Genomics
• Bioinformatics

Background:

• Macroevolutionary key gene identification is a significant challenge in evolutionary biology.
• Understanding gene roles in macroevolution requires advanced analytical methods.

Purpose of the Study:

• To develop a novel computational method for identifying macroevolutionary key genes.
• To investigate the genomic underpinnings of evolutionary divergence in Lepidoptera.

Main Methods:

• Utilized natural language processing (NLP) to create word embedding libraries from genomic data.
• Developed the Integrated Key Gene Method (IKGM) employing domain, kmer, and fused attention mechanisms.
• Applied the IKGM to analyze genomic data from 34 species of butterflies and moths (Lepidoptera).

Main Results:

• Identified specific key genes with high evolutionary weights in Lepidoptera.
• Annotated these genes to functions including circadian rhythms, sensory perception, and behavioral patterns.
• Demonstrated the method's efficacy in pinpointing genes relevant to diurnal and nocturnal adaptations.

Conclusions:

• The IKGM offers a novel genomic-level approach to identify macroevolutionary key genes.
• This method enhances our understanding of microevolutionary mechanisms driving diversification, such as in butterflies and moths.
• Highlights the importance of synergistically acting genes in macroevolutionary processes.