Imbalanced thresholds reduce enzyme prediction accuracy despite high scores
The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction [With Code]
Machine LearningArtificial Intelligence
Summary
Predicting enzyme classes using machine learning is important but tricky because some types are much rarer than others. The authors found that using a standard decision cutoff leads to overly optimistic accuracy numbers while hiding many errors in rare enzyme classes. Their tests showed some classes were guessed too often and others barely guessed at all, even though the model had some ability to tell them apart. They suggest adjusting the decision thresholds specifically for each enzyme class to fix these problems and provide code to do this.
What this means in practice
- •For bioinformatics teams: Improve enzyme function annotation by adjusting prediction thresholds to better detect rare enzyme classes and reduce misleading accuracy scores.
- •For drug discovery software developers: Enhance computational drug discovery tools by applying target-specific threshold optimization to improve enzyme class predictions in imbalanced datasets.
Authors
Bilal Ahmad, Rajed Mehmood
Abstract
Automated prediction of Enzyme Commission (EC) numbers plays a central role in functional annotation and computational drug discovery. However, standard multi-label machine learning pipelines frequently rely on default decision thresholds (t=0.50), assuming balanced prior distributions across target heads. In this study, we present a systematic empirical diagnostic of uncalibrated fixed decision boundaries operating under severe class imbalance across N = 14,096 annotated compounds categorized into six primary EC classes (EC1-EC6). Our results highlight a pronounced Accuracy Paradox: while the multi-label system achieves a deceivingly high mean accuracy of 77.16%, the macro F1-score (0.3976) and macro recall (0.3872) reveal severe predictive breakdown. Majority target classes suffer from hyper-sensitivity and over-prediction, whereas minority classes exhibit sharp recall decay, culminating in a total decision boundary collapse for EC6 (Recall = 0.00%) despite underlying discriminative power (ROC-AUC = 0.5857). Feature correlation analysis further reveals high linear redundancy among topological indices relative to fingerprint density metrics. Ultimately, this diagnostic study demonstrates that standard point predictions mask critical errors in bioinformatics workflows. We establish target-specific threshold optimization and post-hoc conformal calibration as essential, open-source post-processing safeguards for reliable applied machine learning and deep learning architectures.