Variant calling from RNA-Seq data reveals allele-specific differential expression of pathogenic cancer variants.

BACKGROUND

Genetic variants play a pivotal role in the initiation and progression of many diseases, including cancer. Detecting these variants is the first step in understanding their contribution to disease mechanisms. RNA sequencing (RNA-Seq) has become a crucial assay in cancer research, offering insights beyond those provided by DNA sequencing. This study introduces VarRNA, a novel method that utilizes RNA-Seq data to classify single nucleotide variants and insertions/deletions from tumor transcriptomes.

METHODS

VarRNA distinguishes transcriptome variants as germline, somatic, or artifact using a combination of two XGBoost machine learning models. These models were trained and validated using a cohort of pediatric cancer samples with paired tumor and normal DNA exome sequencing data serving as ground truth. We performed additional validation on RNA-Seq data from two distinct cancer datasets, demonstrating that VarRNA outperforms existing RNA variant calling methods.

RESULTS

VarRNA identifies 50% of the variants detected by exome sequencing and detects unique RNA variants absent in paired tumor and normal DNA exome data. Some variants classified by VarRNA exhibit variant allele frequencies distinct from the corresponding DNA exome data. Strikingly, this phenomenon is prevalent in cancer-driving genes, where VarRNA analysis of the RNA-Seq data reveals the variant allele expression as much higher than expected based on the exome sequencing data.

CONCLUSIONS

These findings highlight the potential of RNA-Seq not only to uncover clinically relevant genetic variants but also to offer a deeper understanding of disease-specific expression dynamics that influence cancer pathogenesis, with implications for prognosis and therapeutic strategies.