MSeqDR

MSeqDR performs automated variant interpretation of whole-exome sequencing (WES) datasets for the genetic diagnosis of primary mitochondrial diseases (PMD) by integrating analyses of both nuclear and mitochondrial DNA (mtDNA).

Key Features:

• Dual Genome Analysis: Handles analysis of both nuclear DNA and mtDNA to address the dual genomic origins of PMD.
• Input Requirements: Accepts Variant Call Format (VCF) variant lists, Human Phenotype Ontology (HPO) ID clinical phenotypes, and pedigree files for multi-sample datasets.
• Phenotype Mapping: Maps input clinical phenotypes to HPO terms prior to analysis for phenotype-driven interpretation.
• Efficient Processing: Executes Exomiser in "PASS_ONLY" mode on an 8‑vCPU AWS server and can process up to 100,000 variants in approximately 2 to 20 minutes.
• Variant Ranking and Annotation: Ranks variants by allele frequency, phenotype–gene association, functional impact, and inheritance mode and annotates them using OMIM, ClinVar, dbNSFP, gnomAD, MITOMAP, and MSeqDR.
• Integrated Reports and mtDNA Analyses: Produces Analysis and Integrated reports that include mtDNA-specific outputs such as haplogroup prediction via Phy-Mer, heteroplasmy calculation, and mvTool annotations.

Scientific Applications:

• Pathogenicity Prediction: Implements the Mitochondrial Disease Variant (MDV) classifier using XGBoost trained on over 120 features to predict variant pathogenicity for PMD, reporting >98% accuracy for nuclear gene variants and 94% precision for identifying PMD-causing variants.
• Phenotype-Driven Analysis: Integrates clinical phenotypes as HPO terms to enable phenotype-driven prioritization and interpretation of candidate variants for mitochondrial disease diagnosis.

Methodology:

Clinical phenotypes are standardized to HPO terms and pedigree files are prepared for multi-sample analyses; analysis is executed on VCF inputs using Exomiser in "PASS_ONLY" mode on an 8‑vCPU AWS server to process up to 100,000 variants, with variants ranked by allele frequency, phenotype–gene association, functional impact and inheritance mode and annotated using OMIM, ClinVar, dbNSFP, gnomAD, MITOMAP and MSeqDR; mtDNA analyses include haplogroup prediction with Phy-Mer, heteroplasmy calculation and mvTool annotations, and pathogenicity prediction uses the MDV classifier implemented with XGBoost trained on >120 features.