Frequency Conservation Score (FCS)

Frequency Conservation Score (FCS) predicts and prioritizes pathogenic single-nucleotide variants (SNVs) in nuclear and mitochondrial DNA by integrating evolutionary conservation, population locus variability, and amino-acid physicochemical distance.

Key Features:

• Dual Application: Implements FCS for nuclear DNA and FCSMt for mitochondrial DNA to enable variant assessment across both genomic compartments.
• Random Forest Model: Uses a random forest classifier trained on multiple predictive features to identify pathogenic SNVs.
• Conservation Scores: Incorporates PhastCons and phyloP scores to quantify evolutionary conservation at genomic positions.
• Locus Variability: Integrates locus variability and allele frequency information from the Genome Aggregation Database (gnomAD).
• Physicochemical Distance: Accounts for physicochemical distance of amino-acid substitutions in canonical transcripts as a predictive feature.
• Performance Metrics: Reports an AUC of 0.98 for deleteriousness prediction on an independent validation set, outperforms metaLR, metaSVM, REVEL, DANN, CADD, SIFT, PROVEAN, and FATHMM-MKL, and FCSMt achieves an AUC of 0.92 for pathogenic mitochondrial SNVs.

Scientific Applications:

• Whole Exome Sequencing (WES): Prioritizes pathogenic SNVs detected in WES datasets for downstream interpretation.
• Whole Genome Sequencing (WGS): Prioritizes pathogenic SNVs detected in WGS datasets across nuclear and mitochondrial genomes.
• Clinical Variant Prioritization: Supports identification and ranking of candidate pathogenic variants to inform diagnostic assessment and potential personalized therapeutic considerations.

Methodology:

Integrates PhastCons and phyloP conservation scores, locus variability from gnomAD, and amino-acid physicochemical distance into a random forest classifier trained on these predictors, with a parallel FCSMt implementation for mitochondrial SNVs.