Bioinformatics Tools for Rare Disease Diagnosis

Abstract

Rare diseases collectively affect over 300 million people worldwide, and genomic sequencing has transformed their diagnosis by identifying causal variants in Mendelian disease genes, yet the bioinformatics tools that prioritise candidate variants from the thousands identified per patient vary in diagnostic yield and clinical utility across tool categories. We evaluated 216 bioinformatics rare disease diagnosis programmes across centres in Austria, Germany, and Estonia between 2018 and 2022, spanning five tool categories: phenotype-driven variant prioritisation, gene-phenotype matching algorithms, structural variant detection and annotation, mitochondrial and repeat expansion analysis, and AI-assisted clinical decision support. A Rare Disease Bioinformatics Index (RDBI) was constructed from five sub-scores -- diagnostic yield improvement, candidate list precision, phenotype-genotype concordance, novel gene discovery rate, and time-to-diagnosis reduction -- with weights from regression against sustained adoption into clinical genetics pipelines. RDBI correlated with adoption at r = +0.84 and discriminated adopted from non-adopted tools with an AUC of 0.884. Phenotype-driven prioritisation scored highest (mean RDBI 0.826), while mitochondrial and repeat expansion tools trailed at 0.600. Only 35.6 percent exceeded the 0.75 threshold. Diagnostic yield carried the largest weight (beta = +0.278), followed by candidate list precision (beta = +0.230).