Back

Allele-specific splicing modulates protein isoforms and Alzheimer's risk

King, A. J.; Amoah, K.; Zhang, L.; Bahn, J. H.; Barney, R. M.; Quinones-Valdez, G.; Xiao, X.

2026-02-16 genomics
10.64898/2026.02.13.705796 bioRxiv
Show abstract

Despite growing catalogs of genetic variation linked to human traits and diseases, the functional impact of most genetic variants remains poorly understood. Alternative splicing, particularly in the human brain, represents a key layer of post-transcriptional regulation that may mediate genetic effects on gene expression and protein diversity. In this study, we systematically map allele-specific alternative splicing (ASAS) events in postmortem brain tissues from the Mount Sinai Brain Bank cohort, identifying hundreds of genetically regulated splicing events across four brain regions. Using a concordance-based method, we nominate over 500 putative functional SNPs associated with ASAS, many of which overlap splicing QTLs (sQTLs), RNA-binding protein binding sites, and GWAS loci for Alzheimers disease (AD), brain traits, and immune phenotypes. ASAS events are enriched in genes involved in mitochondrial function and frequently occur in 5' untranslated regions (5' UTRs), where they are associated with protein quantitative trait loci (pQTLs), alternative start codons, and isoform-specific domain changes - highlighting an underappreciated mechanism through which noncoding variants can influence translation and proteome complexity. Importantly, we also identify a subset of ASAS events exhibiting disease-specific splicing patterns in AD brains, including functional SNPs with opposing splicing effects between AD and control groups in genes implicated in mitochondrial function and neuronal signaling. Together, our results provide a brain-specific, splicing-resolved map of regulatory variation and uncover novel mechanisms linking genetic variation to transcript and protein-level changes in AD. This work highlights the importance of allele-specific splicing analysis for interpreting noncoding variation in complex human disorders.

Published in Genome Research (predicted rank #1) · training set

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"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.