RNA-SSNV

RNA-SSNV identifies somatic single nucleotide variants (SSNVs) from bulk RNA sequencing (RNA-seq) of tumor samples and distinguishes true expressed somatic mutations from RNA-specific artifacts for cancer genomics applications.

Key Features:

• Multi-Filtering Strategy: Employs a multi-filtering approach to remove RNA-editing events, reverse transcription artifacts, alignment gaps, and sequencing noise from variant calls.
• Machine Learning Classification Model: Implements a machine-learning classifier trained on curated features to improve classification, achieving test precision-recall rates of 0.880–0.884.
• Robust Validation: Validated on three adult-based TCGA datasets with a precision-recall AUC of 0.94.
• Variant Allele Fraction (VAF) Analysis: Enables analysis of VAF to assess subclonal selection and the expression levels of detected variants.
• Expressed Mutation Prioritization: Prioritizes somatic mutations with higher functional impact and therapeutic relevance in known driver genes.

Scientific Applications:

• Cancer driver mutation identification: Identification and prioritization of active driver mutations from tumor RNA-seq to inform studies of tumor biology and potential therapeutic targets.
• Tumor evolution and subclonal analysis: Analysis of VAF to infer evolutionary selection advantages of subclonal expressed mutations.
• Complementing DNA-based analyses: Detection of expressed somatic mutations that may be missed by DNA-only sequencing to provide a more complete view of carcinogenic mechanisms.

Methodology:

Applies multi-filtering to remove RNA-specific artifacts and a machine-learning classifier trained on curated features; performance was evaluated on three adult-based TCGA datasets reporting a precision-recall AUC of 0.94 and test precision-recall rates of 0.880–0.884.