Machine learning models identify gene predictors of waggle dance behaviour in honeybees

Area of Science:

• * Neurogenomics and computational biology.
• * Analysis of complex behavioral phenotypes.
• * Application of machine learning in biological research.

Background:

• * Understanding the genetic basis of complex behaviors is challenging due to multiple contributing factors.
• * Traditional neurogenomic analyses face limitations, notably the
• when analyzing gene expression data with few replicates.
• * The honeybee waggle dance serves as a model for studying the genomic underpinnings of social behavior.

Purpose of the Study:

• * To implement and evaluate machine learning algorithms for transcriptomic analysis.
• * To identify genes associated with the honeybee waggle dance using computational approaches.
• * To complement traditional differential gene expression analysis with machine learning techniques.

Main Methods:

• * Implementation of three supervised machine learning algorithms: Random Forests, Lasso and Elastic net Regularized Generalized Linear Model, and Support Vector Machine.
• * Application of these algorithms to characterize transcriptomic patterns related to the honeybee waggle dance.
• * Comparison of machine learning results with traditional differential gene expression analysis.

Main Results:

• * Machine learning approaches successfully characterized transcriptomic patterns associated with the honeybee waggle dance.
• * The study identified two promising candidate genes, "boss" and "hnRNP A1," involved in the neural regulation of this behavior.
• * Machine learning provided a valuable complement to traditional gene expression analysis methods.

Conclusions:

• * Machine learning is a powerful tool for analyzing transcriptomics data and identifying candidate genes for complex social behaviors.
• * The identified genes "boss" and "hnRNP A1" warrant further investigation for their role in the honeybee waggle dance.
• * This ML-based approach offers advantages over traditional methods and has broad applications in evolutionary ecology and behavioral genomics.