Multiple sclerosis (MS) is an autoimmune disorder with complex genetic and environmental associations in its disease pathophysiology. MS prevalence is characterised by a complex geographical profile, with both populations of European ancestry and populations at higher latitude experiencing higher prevalence of MS. Our hypothesis is that this geographical pattern of MS prevalence is likely to have been shaped by natural selection, and specifically local adaptation.
Landscape genomics examines the relationship between genomics and spatial information on environmental factors to identify candidate genes undergoing local adaptation. While predominantly used in ecological studies, landscape genomics holds promise in human genetics, particularly in understanding diseases like MS, which exhibits a strong latitudinal gradient in its prevalence.
We conducted a gene–environment analysis using data from the UK Biobank on a genomic wide scale, focusing on the association with varying UV levels to detect genomic regions that show signals of selection influenced by UV exposure. Genotyping and associated metadata, including blood serum vitamin D levels, latitude, and longitude were obtained from UK Biobank. UV levels were obtained from WorldClim dataset. Logistic regression models were run genome wide for 20,000 individuals to identify genetic regions associated with MS and exhibiting selection signals. Subsequently, an interpretable machine learning approach was applied to delve deeper into these identified genetic regions, providing insights into the relationship between single nucleotide polymorphisms (SNPs) and environmental variables. Ongoing analysis aims to determine variable importance related to the genomic regions of interest and identify SNPs undergoing local adaptation.