RF-DisLncPred: Utilizing Random Forest for the Prediction of Disease-Linked Long Non-Coding RNAs

Previously dismissed as mere transcriptome byproducts, long noncoding RNAs (lncRNAs) are today acknowledged to act as crucial master regulators over various biological processes.They are required for processes such as genomic imprinting, cell differentiation, and RNA splicing.Numerous diseases have been linked to their malfunctions.We now have the impetus to investigate their therapeutic potential in much more depth.Crucially, and hopefully, their potential as antipathogenic agents will be realized in the future.In this study, we present RF-DisLncPred, a creative computational method for improving disease-associated lncRNA detection.This approach synthesizes diverse features, utilizes the chi-square method for optimal feature selection, and employs a random forest classifier for effective prediction.Our process was strictly validated using the 10-fold cross-validation method to test its robustness.It showed superior performance, with 90.86%accuracy, 87.32% sensitivity, 94.21% precision, and an F-score of 90.54%.Compared to existing advanced methods, comparative studies have shown that RF-DisLncPred performs the best in terms of F-score.This study addresses the challenges that have plagued prior research with respect to sample redundancy and class imbalance.The accurate combination of different characteristics or the identification of meaningful descriptors via the chi-square method can greatly improve its ability to predict disease-related lncRNAs.Extensive experiments demonstrate that our findings can largely advance the state of the art in machine learning-based models for lncRNAdisease association, showcasing the exciting potential for sequence variations within lncRNAs to be preserved when making disease predictions.