The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction [With Code]
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.
Related
- Diagnostic accuracy of Mycobacterium tuberculosis antigen-induced interferon-γ-inducible protein 10 for latent tuberculosis infection among high-risk groups: a systematic review and meta-analysis
- Beyond prediction accuracy by integrating GWAASB into wheat stability assessment across environments in Saudi Arabia
- Does visual attention to diagnostic facial features predict eyewitness identification accuracy? Implications for diagnostic feature detection theory
- Diagnostic accuracy of Neutrophil Gelatinase-Associated Lipocalin (NGAL) for inflammatory bowel disease: a systematic review and bivariate diagnostic meta-analysis.
- Artificial intelligence and digital health in bronchoscopy: advances in diagnostic accuracy, real-time navigation and procedural skills training
BibTeX
@misc{ahmad2026accuracy,
title = {{The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction [With Code]}},
author = {Bilal Ahmad and Rajed Mehmood},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.07897},
eprint = {2609.07897},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.07897},
}
RIS
TY - UNPB
TI - The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction [With Code]
AU - Bilal Ahmad
AU - Rajed Mehmood
PY - 2026
DA - 2026/09/07
PB - arXiv
DO - 10.48550/arxiv.2609.07897
UR - https://arxiv.org/abs/2609.07897
AB - 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.
ER -