Cancer gene variant identification and functional interpretation using long-read RNA sequencing with FLAIR3

Although the impact of single nucleotide variants (SNVs) and changes in transcription and RNA processing are often analyzed separately, a comprehensive analysis facilitates a complete understanding of how cancer gene alterations impact oncogenesis. In traditional short-read RNA sequencing, phasing of alternative exons and cancer variants is lost because the read lengths are much shorter than typical mRNA transcripts (average > 1kb). Here, we show that long-read RNA-seq (lrRNA-seq) can identify full-length transcript isoforms on which variants are expressed, which can be used to more accurately identify the functional impact of oncogenic variants. We developed FLAIR3, which performs an integrated analysis of SNVs, gene fusions, and alternative splicing using lrRNA-seq and predicts functional changes to the amino acid sequence. We performed ONT lrRNA-seq on three osteosarcoma cell lines and PacBio lrRNA-seq on paired normal and tumor tissue from two lung adenocarcinomas. We then used FLAIR3 to identify cancer driver variants and to determine how splicing modulates their expression and function. In the osteosarcoma samples, FLAIR3 revealed alternatively spliced gene fusions in cancer driver genes and TP53 gene fusions with intergenic regions, predicted to cause TP53 truncations. In the lung adenocarcinomas, FLAIR3 revealed isoform-biased expression of oncogenic BRAF V600E . Through an isoform-specific analysis of somatic SNVs in CDKN2A, we found that TP53 loss significantly co-occurs with CDKN2A missense or nonsense variants of the p16 isoform, but not with CDKN2A deep deletion. Damaging variants in the p16 isoform would not have the same damaging effects in p14 isoform, which functions through TP53; therefore, TP53 loss would be necessary to have complete loss of CDKN2A functions. A deep deletion of CDKN2A removes both p16 and p14 isoform function and would not need to have additional TP53 loss. These findings reveal how alternative splicing interacts with and modulates the function of oncogenic variants.