Advancing Enzyme-Based Detoxification Prediction with ToxZyme: An Ensemble Machine Learning Approach

Toxins, Год журнала: 2025, Номер 17(4), С. 171 - 171

Опубликована: Апрель 1, 2025

The aaccurate prediction of enzymes with environment detoxification functions is crucial, not only to achieve a better understanding bioremediation strategies, but also alleviate environmental pollution. In the present study, novel machine learning model was introduced which classifies by their toxin degradation ability. this model, two different sets data were used include that can catalyze as positive dataset and non-toxin-degrading negative dataset. Further, comparison multiple classifiers performed find best Random Forest (RF) classifier selected due its strong performance. To enhance accuracy, we combined RF Deep Neural Network (DNN), forming an ensemble effectively integrated both techniques. This combination achieved 95% precision, surpassing individual models. Our ensures high accuracy reliably differentiates toxin-degrading from non-degrading ones. study highlights power combining classical deep advance prediction. represents significant step in enzyme classification serves valuable resource for biotechnology, food nutrition, health applications.

Язык: Английский

Machine learning based boosting models for predicting flexural strength of steel fiber reinforced concrete

Asian Journal of Civil Engineering, Год журнала: 2025, Номер unknown

Опубликована: Июнь 3, 2025

Язык: Английский