Deep learning of 777 K bulk transcriptomes reveals human-mouse gene conservation beyond DNA sequence similarity

Area of Science:

• Genomics
• Comparative Biology
• Bioinformatics

Background:

• Mice are common biomedical research models, but discrepancies arise between mouse and human study results despite DNA sequence similarities.
• Understanding gene relationships across species is vital for accurate translational research.

Purpose of the Study:

• To investigate human-mouse gene relationships using gene expression patterns.
• To identify genes with differing functional associations between humans and mice.
• To provide a resource for assessing mouse model suitability for human gene studies.

Main Methods:

• Utilized the Transformer-based GeneRAIN model on 777,000 human and mouse bulk RNA sequencing samples.
• Generated RNA-based gene representations to compare homologous genes.
• Analyzed DNA sequence similarity alongside RNA characteristics.

Main Results:

• Identified 2,407 human-mouse homologous genes with high DNA similarity but distinct RNA profiles, indicating potential differences in disease associations.
• Discovered 3,070 genes with low similarity at both DNA and RNA levels, suggesting a high risk of cross-species discrepancies.
• Identified long noncoding RNA homologs based on cross-species RNA conservation.

Conclusions:

• Gene expression patterns offer critical insights into human-mouse gene relationships beyond DNA sequence.
• The findings highlight potential pitfalls in using mouse models for specific human genes.
• This study provides a valuable resource for researchers to evaluate the translational relevance of mouse models.