Genetic Variant Classification Through Decision Tree Analysis for Enhanced Genomic Understanding

Summary

Genomic research has grown immensely, leading to a surge in the discovery of genetic variations. Efficient methods are needed to classify these variants based on their functional significance. This study explores using decision tree analysis as a powerful tool to classify genetic variants, aiming to deepen our grasp of the genomic landscape. We use a comprehensive dataset containing various genetic variants from different genomic databases. We process the data using bioinformatics tools and computational algorithms to identify essential features and build a decision tree model. The model is trained and tested using datasets with known variant classifications to ensure accuracy. The decision tree model effectively sorts genetic variants into meaningful categories like pathogenic, benign, and uncertain. The simplicity of decision trees helps identify key genomic features affecting the classification, shedding light on biological mechanisms. Additionally, we examine how well the decision tree model works across diverse genomic scenarios, such as different genes and functional regions. The results show the model's reliability and potential for broad use in variant classification. This research contributes to ongoing efforts to simplify and automate genetic variant classification, tackling the challenges posed by the growing genomic data landscape. The proposed decision tree method improves the interpretation of variants and supports the larger goal of advancing personalized medicine and genomic diagnostics.The random forest algorithm yielded precision values of 0.87 and 0.40 and recall values of 0.63 and 0.72, with corresponding F1‐scores of 0.73 and 0.51, achieving an accuracy of 65.34%. The gradient descent model displayed 88% accuracy for class 0 and 43% for class 1, with F1‐scores of 0.77 and 0.54, and an overall accuracy of 68.97%. The decision tree classification demonstrated precision of 0.85 and 0.41, recall of 0.68 and 0.65, resulting in F1‐scores of 0.75 and 0.50, and an accuracy of 67.1%. The final model achieved an average recall of 0.69, effectively identifying inconsistent assessments of genetic variations.