A basic framework governing splice-site choice in eukaryotes
Dent, C. I.; Prodic, S.; Balakrishnan, A.; Georges, J.; Chhabra, A.; Mukherjee, S.; Coutts, J. P.; Gitonobel, M.; Sarwade, R. D.; Rosenbluh, J.; D'Amato, M.; Das, P. P.; Guo, Y.; Fournier-Level, A.; Burke, R.; Sureshkumar, S.; Powell, D.; Balasubramanian, S.
Show abstract
Changes in splicing are observed between cells, tissues, organs, individuals, and species. These changes can mediate phenotypic variation ranging from flowering time differences in plants to genetic diseases in humans. However, the genomic determinants of splicing variation are largely unknown. Here, we quantified the usage of individual splice-sites and uncover extensive variation between individuals (genotypes) in Arabidopsis, Drosophila and Humans. We used this robust quantitative measure as a phenotype and mapped variation in splice-site usage using Genome-Wide Association Studies (GWAS). By carrying out more than 130,000 GWAS with splice-site usage phenotypes, we reveal genetic variants associated with differential usage of specific splice-sites. Our analysis conclusively shows that most of the common, genetically controlled variation in splicing is cis and there are no major trans hotspots in any of the three analyzed species. High-resolution mapping allowed us to determine genome-wide patterns that govern splice-site choice. We reveal that the variability in the intronic hexamer sequence (GT[N]4 or [N]4AG) differentiates intrinsic splice-site strength and is among the primary determinants of splice-site choice. Experimental analysis validates the primary role for intronic hexamer sequences in conferring splice-site decisions. Transcriptome analyses in diverse species across the tree of life reveals that hexamer rankings explains splice-site choices from yeast to plants to humans, forming the basic framework of the splicing code in eukaryotes.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.