Predicting gene expression divergence between single-copy orthologs in two species

Area of Science:

• Evolutionary biology
• Genomics
• Bioinformatics

Background:

• Predicting gene expression divergence is key to understanding the evolution of biological functions and traits.
• Existing methods are limited to duplicate genes or require data from multiple species.

Purpose of the Study:

• Introduce PiXi, a novel machine learning framework for predicting gene expression divergence between single-copy orthologs in two species.
• Provide a tool to analyze gene expression evolution and identify origins of novel phenotypes.

Main Methods:

• PiXi models gene expression evolution using an Ornstein-Uhlenbeck process.
• It employs multi-layer neural networks, random forests, and support vector machines for prediction.
• The framework predicts conserved or diverged expression and expression optima for ortholog pairs.

Main Results:

• PiXi demonstrates high accuracy and power in predicting gene expression divergence and optima for single-copy orthologs.
• A multi-layer neural network architecture achieved the best overall performance.
• Empirical data from Drosophila showed approximately 23% of relocated genes diverged in expression, with some linked to mitochondrial energy production.

Conclusions:

• PiXi offers a robust toolkit for predicting gene expression divergence between single-copy orthologs in two species.
• The findings suggest that positional gene relocation can influence expression and potentially lead to new phenotypes, as seen in Drosophila's energy production pathways.