Integrating machine learning and functional genomics to study cross-species gene regulatory evolution

Area of Science:

• Evolutionary biology
• Genomics
• Gene regulation

Background:

• Understanding species phenotypic differences requires identifying genetic underpinnings.
• Mammalian genomes reveal many sequence differences, mostly in noncoding regions.
• Cis-regulatory elements (CREs) control gene expression but are complex and hard to study.

Purpose of the Study:

• To review recent advances in studying gene regulatory evolution.
• To highlight experimental and computational strategies for understanding CRE function across species.

Main Methods:

• Comparative genomics to identify sequence differences.
• Functional profiling across species to assess gene expression and CRE activity.
• High-throughput perturbation assays to test CRE function.
• Machine learning to predict CRE activity from DNA sequences.

Main Results:

• Advances allow cataloging cross-species differences in gene expression and CRE function.
• Machine learning models are emerging to predict CRE activity from sequence data alone.

Conclusions:

• Integrating experimental and computational approaches is key to deciphering gene regulatory evolution.
• New strategies are improving our ability to link genetic changes in CREs to functional outcomes across species.