Scholarlink

<p>Malaria is a significant global health challenge, causing high morbidity and mortality. The rise of drug resistance highlights the urgent need for new antimalarial agents. This study focuses on predictive modeling of 104 <i>Plasmodium falciparum</i> protein kinase 6 (PfPK6) inhibitors, employing a range of machine learning techniques to develop ensemble regression and classification models. Molecular descriptors were refined using classification and regression trees (CART) to identify the most relevant features. Six machine learning algorithms (Random Forest (RF), Relevance Vector Machine (RVM), Support Vector Machine (SVM), Cubist, Artificial Neural Networks (ANN), and XGBoost) were utilized to construct regression models. The consensus model demonstrated superior predictive performance, achieving R<sup>2</sup><sub>Test</sub> = 0.94, SE<sub>Test</sub> = 0.20, Q<sup>2</sup><sub>CV</sub> = 0.90, and SE<sub>CV</sub> = 0.25, outperforming individual models. For classification tasks, five algorithms were evaluated and a majority voting approach yielded an accuracy of 91% and a sensitivity of 93%. The robustness of the models was confirmed through applicability domain analysis (96% coverage) and y-randomization tests, ensuring that the predictive outcomes were not due to chance correlations. This study highlights the effectiveness of ensemble machine learning approaches in predictive modeling and provides critical insights for the rational design of novel PfPK6 inhibitors.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Advancing antimalarial drug discovery: ensemble machine learning models for predicting PfPK6 inhibitor activity

  • 期刊论文

  • 发表日期：

    2025年4月22日

查看全文
  • Maryam Gholami,
  • Mohammad Asadollahi-Baboli

摘要

Malaria is a significant global health challenge, causing high morbidity and mortality. The rise of drug resistance highlights the urgent need for new antimalarial agents. This study focuses on predictive modeling of 104 Plasmodium falciparum protein kinase 6 (PfPK6) inhibitors, employing a range of machine learning techniques to develop ensemble regression and classification models. Molecular descriptors were refined using classification and regression trees (CART) to identify the most relevant features. Six machine learning algorithms (Random Forest (RF), Relevance Vector Machine (RVM), Support Vector Machine (SVM), Cubist, Artificial Neural Networks (ANN), and XGBoost) were utilized to construct regression models. The consensus model demonstrated superior predictive performance, achieving R2Test = 0.94, SETest = 0.20, Q2CV = 0.90, and SECV = 0.25, outperforming individual models. For classification tasks, five algorithms were evaluated and a majority voting approach yielded an accuracy of 91% and a sensitivity of 93%. The robustness of the models was confirmed through applicability domain analysis (96% coverage) and y-randomization tests, ensuring that the predictive outcomes were not due to chance correlations. This study highlights the effectiveness of ensemble machine learning approaches in predictive modeling and provides critical insights for the rational design of novel PfPK6 inhibitors.