Document Type
Article
Publication Date
Winter 1-22-2024
Abstract
The most prevalent kind of genetic variants in humans are non-synonymous single nucleotide variants (nsSNVs). Several prediction tools have been launched to forecast the effect of amino acid substitutes on human protein function. These tools sort variants as pathogenic or neutral. We developed an Integrated Rules Classifier (Integration Score through JRip “ISTJRip”), which integrates the four individual tools that are publicly available; iFish, Mutation Assessor, FATHMM, and SIFT-based on the JRip machine learning technique. Additionally, we compared the ISTJRip approach with the other three created integration classifiers; Integration Score through J48 “ISTJ48”, Integration Score through RF “ISTRF”, and Integration Score through SVM “ISTSVM” using a VaribenchSelectedPure dataset character from the standard dataset “Varibench”. The proposed integrated rules classifier “ISTJRip” and the other three integration classifiers, ISTJ48, ISTRF, and ISTSVM register 92.41 %, 92.26 %, 91.70 %, and 90.62 % ACC on VaribenchSelectedPure, respectively. Finally, we demonstrated that the integrated rules classifier outperforms other integration classifiers and highlights the benefits of JRip machine learning technique in the integration process for multiple tools.
Recommended Citation
Hosseny, Ahmed Barakat; Hassan, Marwa Said; Shalan, A A.; Khamis, Shymaa; and Dessouky, M I., "Integrated rules classifier for predicting pathogenic non-synonymous single nucleotide variants in human" (2024). Basic Science Engineering. 138.
https://buescholar.bue.edu.eg/basic_sci_eng/138