Quinones-Valdez, G.; Amoah, K.; Xiao, X.

Identification of cis-directed splicing events is instrumental in elucidating the genetic basis of splicing changes. In this study, we introduce isoLASER, a computational method leveraging long-read RNA-seq data to distinguish splicing events predominantly influenced by cis- or trans-regulatory mechanisms. IsoLASER computes the adjusted mutual information between haplotypes and exonic segments, and sensitively identifies allele-specific splicing events compared to transcript-level approaches. Using isoLASER, we identified 2,047 and 4,679 unique exonic regions exhibiting cis-directed splicing in human and mouse data respectively. Moreover, applied to a small number of samples jointly, isoLASER reveals splicing-associated variants that were only identifiable using large sample cohorts in previous association studies. Importantly, isoLASER enables HLA typing using long-read RNA-seq alone and uncovers allele-specific splicing of HLA genes. In brain tissues, we identified variants linked to splicing in Alzheimer`s disease-relevant genes such as MAPT, BIN1 and MS4A6A. Together, we establish a framework for identifying genetically and non-genetically driven splicing and associated variants, particularly in cohorts with limited sample sizes.