Líffræðifélag Íslands - biologia.is
Líffræðiráðstefnan 2021

Erindi/veggspjald / Talk/poster V47

Using machine learning to predict the genomic structure of gene expression.

Höfundar / Authors: Oren Raz(1), Quentin Horta-Lacueva (1), Kalina Kapralova (1)

Starfsvettvangur / Affiliations: 1. Háskóli Íslands

Kynnir / Presenter: Oren Raz

Machine learning is contributing to major advances in molecular biology, specifically in gene expression studies. In this project, we use machine learning to characterise spatial patterns of gene expression in the genome of diverging populations. We used two Arctic charr morphs (Salvelinus alpinus) that have diverged to occupy benthic and pelagic feeding habitats in lake Thingvallavatn. Embryos of both charr morphs were reared in common garden conditions and sampled for RNA-sequencing. We identified genes that were differentially expressed between the two morphs and extracted their genomic coordinates from the reference charr genome. We are now using Random Forest algorithms to estimate whether the differentially expressed genes cluster across the charr genome to a greater extent than expected by chance. We also developed tools to visualize the distribution of the identified genomic regions with coordinated gene expression. The information related to these regions could later be used for functional analyses (GO ontology) and coupled with other genomic approaches (genomic landscape, genetic map, inversion map) to unravel how the influence of genetic/-omic architecture on gene expression determine phenotypic divergence.