rfPred

rfPred predicts variant pathogenicity by integrating SIFT, Polyphen2, LRT, PhyloP and MutationTaster scores from dbNSFP and applying a random forest meta-classifier to prioritize non-synonymous single-nucleotide variants.

Key Features:

• Multi-Score Integration: Combines SIFT, Polyphen2, LRT, PhyloP and MutationTaster scores from dbNSFP to assess variant impact.
• Machine Learning Meta-Score: Uses a random forest to generate a meta-score with reported higher accuracy than individual scores or CADD on validation datasets.
• Pre-Computed Human Exome Scores: Provides scores for all non-synonymous SNPs in the human exome.

Scientific Applications:

• Genomic Variant Prioritization: Identifies deleterious non-synonymous variants for experimental validation and prioritization in genomic studies.

Methodology:

Trains a random forest on 61,500 non-synonymous SNPs using integrated scores (SIFT, Polyphen2, LRT, PhyloP, MutationTaster from dbNSFP) to generate a meta-score for scalable analysis of next-generation sequencing data.