MALER

Machine learning (ML) methods have accelerated discoveries in biological and biomedical research, especially in the analysis of big omics data. However, biologists often lack knowledge of ML theories and programming skills to easily apply ML in their research. The lack of a one-stop web server to build, evaluate, and compare the performance of different ML models for biological data has been a major obstacle for biologists to apply ML widely.

We developed MALER (Machine Learning Evaluation and Reporting server), a guided web application that helps biologists apply ML models to biological and biomedical data without writing code. MALER provides modules for (i) binary and (ii) multi-class classification, and (iii) regression and (iv) survival analysis that allow users to build, quantitatively evaluate, and compare multiple ML models with task-appropriate metrics. The reviewed local workflow refits learned transforms within model-search training folds. One-click compares per-family outer development estimates; choosing a family from those scores can make the winning mean optimistic. Only the chosen family is retuned and fitted on development data, then its fitted configuration is locked before held-out evaluation. Predict reuses its signed pipeline without retraining. Grouped and explicit-cutoff FSS/BSS are rejected; the sequential selectors' additional internal scores do not refit enclosing-fold preprocessing.

In the ANALYSIS page, users can upload their omics profile dataset and the corresponding labels to build and compare their custom ML models with step by step guidance. In the PREDICT page, users can upload their ML model and a omics profile with unknown labels to predict outcomes. You can find detailed instructions on how to use MALER on the HELP page.