Staem5: A novel computational approach for accurate prediction of m5C site

Abstract

5-Methylcytosine (m5C) is an important post-transcriptional modification that has been extensively found in multiple types of RNAs. Many studies have shown that m5C plays vital roles in many biological functions, such as RNA structure stability and metabolism. Computational approaches act as an efficient way to identify m5C sites from high-throughput RNA sequence data and help interpret the functional mechanism of this important modification. This study proposed a novel species-specific computational approach, Staem5, to accurately predict RNA m5C sites in Mus musculus and Arabidopsis thaliana . Staem5 was developed by employing feature fusion tactics to leverage informatic sequence profiles, and a stacking ensemble learning framework combined five popular machine learning algorithms. Extensive benchmarking tests demonstrated that Staem5 outperformed state-of-the-art approaches in both cross-validation and independent tests. We provide the source code of Staem5, which is publicly available at https://github.com/Cxd-626/Staem5.git .

Biological roles of RNA m5C modification and its implications in Cancer immunotherapy

“…In bisulfite sequencing, Bisulfite treatment can convert the unmethylated C bases in the genome into U, which becomes T after PCR amplification to distinguish them from the original C bases with methylation modifications, and then combined with high-throughput sequencing technology, a genome-wide DNA methylation map with single-base resolution can be drawn [33]. Moreover, a novel species-specific computational approach, Staem5, to accurately predict RNA m 5 C sites in Mus musculus and Arabidopsis thaliana was recently developed [34].…”

m5Cpred-XS: A New Method for Predicting RNA m5C Sites Based on XGBoost and SHAP

“…To further evaluate the generalization of our models, the predictive results of our models on the independent test sets were compared with other existing methods, iRNA-m5C ( Lv et al, 2020 ), m5CPred-SVM ( Chen Xiao et al, 2020 ), RNAm5Cfinder ( Li et al, 2018 ), iRNAm5C-PseDNC ( Qiu et al, 2017 ), RNAm5CPred ( Fang et al, 2019 ), PEA-m5C ( Song et al, 2018 ), and Staem5 ( Chai et al, 2021b ). However, not all of these methods can predict m5C sites in all three species.…”

“…Furthermore, Visentini et al (2016) first sorted the features according to the F-score obtained in the eXtreme gradient boosting (XGBoost) ( Chen, 2016 ) package and then selected the top 50 features based on the incremental feature selection (IFS) strategy as the optimal feature subset. To reduce the dimension of features, Chai et al (2021a) proposed an efficient m5C sites prediction approach, Staem5, based on features selected by F-score. The SHapley Additive ExPlanations (SHAP) ( Wang and Gribskov 2019 ; Bi et al, 2020 ) method, which can interpret the importance of features, is another effective method for selecting relevant features.…”

MLACP 2.0: An updated machine learning tool for anticancer peptide prediction

“…Construction of baseline models: We utilized seven different classifiers (RF, ERT, SVM GB, AB, LGB, and XGB) that have been extensively applied in Bioinformatics and computational biology [32] , [33] , [34] , [35] , [36] , [37] . For each classifier, there are a set of hyperparameters that determine the performance of the model during cross-validation.…”