DTreePred

DTreePred predicts pathogenicity of genomic variants using a machine learning-based pathogenicity model that integrates data from public databases and multiple prediction algorithms to support clinical decision-making.

Key Features:

• Machine learning pathogenicity model: DTreePred employs a decision tree algorithm and 15 machine learning classifiers alongside classical predictors for variant pathogenicity assessment.
• Public database integration: The tool connects public databases with diverse prediction algorithms to streamline variant analysis.
• Cross-platform accessibility: DTreePred functions as both a cross-platform mobile application and an online visualization tool.
• High accuracy performance: An integrity test with 200 known mutations yielded 97% accuracy, and a case study with nucleotide variants from oncogenes and suppressor genes classified in ClinVar as inexact data successfully revealed pathogenicity of more than 95% of variants.

Scientific Applications:

• Clinical variant analysis: DTreePred supports decision-making in clinical practice by reducing uncertainty in pathogenicity assessments of nucleotide variants.
• Cancer genomics: The tool categorizes nucleotide variants from lists of oncogenes and suppressor genes to reveal variant pathogenicity.
• Underrepresented population studies: DTreePred addresses the lack of representativeness of single nucleotide variants in public databases and low sequencing rates in underrepresented populations.
• Precision medicine: Improving accuracy of pathogenicity assessments increases precision of medical diagnoses and treatments.

Methodology:

DTreePred utilizes a decision tree algorithm and 15 machine learning classifiers alongside classical predictors, integrating information from various public databases and prediction techniques.