Off the deep end: What can deep learning do for the gene expression field?

Area of Science:

• Genetics
• Evolutionary Biology
• Bioinformatics

Background:

• The fifth biennial symposium on Evolution and Core Processes in Gene Regulation convened in July 2022, focusing on advancements in understanding gene regulation.
• Experts in gene regulation and evolutionary biology discussed key areas including enhancer evolution, the cis-regulatory code, and regulatory variation.

Purpose of the Study:

• To explore the application of deep learning (DL) in deciphering DNA sequence information for gene regulation.
• To highlight how DL can identify impactful genetic variants and elucidate the mechanistic underpinnings of the cis-regulatory code.

Main Methods:

• The symposium focused on the application of deep learning (DL), a machine learning technique utilizing neural networks.
• DL models were trained to predict genomic data from DNA sequences, enabling high prediction accuracy.

Main Results:

• High prediction accuracy of DL models facilitates the identification of impactful genetic variants within and across species.
• Extracted sequence rules from DL models offer insights into the mechanistic basis of the cis-regulatory code.

Conclusions:

• Deep learning is a powerful tool for analyzing DNA sequence information in gene regulation research.
• DL advancements aid in understanding evolutionary processes and identifying functional genetic elements.