Direct long-read RNA sequencing uncovers functional variation affecting transcript production and RNA modifications

The production of multiple transcripts per gene is a process regulated by inherited genetic variants and epitranscriptomic modifications, and plays a prominent role in modulating complex traits and diseases. To simultaneously characterize the effect of genetic variants on transcript abundance and N6-methyladenosine (m⁶A) modifications, we produced long-read native poly(A) RNA-seq data for 60 genetically different lymphoblastoid cell lines (LCLs) from the 1000 Genomes/Geuvadis project. We identified a high diversity of both annotated (40.2%) and unannotated (59.8%) transcripts, with only a small proportion expressed across individuals (35% and 18.4%, respectively). In a genome-wide genetic analysis on transcripts, we identified 87 trQTLs (transcript quantitative trait loci), of which 55 were not detected as eQTLs using a larger published short-read RNAseq dataset (317 samples). A population wide characterization of m⁶A methylation DRACH motifs (canonical m⁶A consensus sequence) identified an average of 4.04 m⁶A modifications on 4,325 genes. Genetic association analysis of highly variable modifications from 1,272 genes identified m⁶A modification quantitative trait loci (m⁶A-QTLs) for 5 transcripts. Even with a limited number of genetic associations, colocalization analysis of trQTLs and m⁶A-QTLs identified 8 candidate transcripts mediating GWAS traits, with approximately 75% of the colocalized trQTLs implicating novel risk transcripts, and an enrichment of GWAS lead variants involved in trQTLs. Overall, the simultaneous characterization of transcripts and post-transcriptional modifications identified genetic effects on transcription often missed when using other sequencing technologies.