More and more researches are using omics technologies, such as genome, transcriptome, proteome, and metabolome to uncover biomarker discovery in cohort studies. Machine learning methods are extensively employed to identify significant features from high-throughput omics data and construct models for clinical applications such as early screening, risk prediction, etc. The implementation of machine learning is limited due to the requirement of a specialized professional experience. Therefore, we have developed Omics Machine Learning (OML), a comprehensive and user-friendly online platform for supervised machine learning tasks, including binary classification, multi-classification, and regression. The classification models comprise a set of 10 commonly used machine learning algorithms, while the regression models encompass 8 commonly used algorithms. OML incorporates three data set splitting methods (cross-validation, percentage splitting, and custom training/validation sets) to cater different research design needs. Furthermore, OML offers thorough analysis of model performance interpretation outcomes. By utilizing four publicly accessible large cohort datasets, re-processing in OML with detailed parameter settings, comparing comprehensive interpretations for model performance, OML yield results that are equivalent to or surpassing those in original research. The OML platform is free availability at https://omia.untangledbio.com/oml/.
Please send any comment, suggestion or question you may have to the author (Ms. Zhang), email: 2685743859@qq.com.