Attentive interpretable tabular learning (TabNet) for miRNA expression classification in cervical cancer and Alzheimer's disease
- 1Department of Computer Engineering, Hasan Kalyoncu University, Gaziantep, 27020 Turkey
- 2Department of Computer Engineering, Gaziantep University, Gaziantep, 27310 Turkey
Abstract
Background and Objective: The classification of RNA sequencing (RNA-seq) data presents a fundamental challenge in computational biomedicine: raw read counts are high-dimensional, over-dispersed, and heteroscedastic, requiring careful preprocessing before machine learning models can be applied. Existing deep learning approaches for RNA-seq classification have largely overlooked the tabular nature of genomic expression data and have offered limited embedded interpretability. This study proposes TabNet, an attentive interpretable tabular learning architecture, as a framework for miRNA-based disease classification and evaluates its performance and interpretability against three established baseline models.
Methods: Two publicly available miRNA expression datasets were examined: a cervical cancer cohort (58 samples, 714 miRNAs) and an Alzheimer's disease cohort (70 samples, 503 miRNAs). Raw count data were preprocessed following the MLSeq framework, encompassing DESeq median-ratio normalisation, near-zero variance filtering, supervised gene ranking based on DESeq differential expression analysis, and the DESeq variance-stabilising transformation (vst). Gene subsets corresponding to 5%, 10%, and 30% of the post-filtering feature space were evaluated. TabNet was benchmarked against eight baselines spanning four learning paradigms: a 1D Convolutional Neural Network (CNN), a Long Short-Term Memory network (LSTM), Extreme Gradient Boosting (XGBoost), Random Forest, ElasticNet-regularised logistic regression, a Multi-Layer Perceptron (MLP), the prior-fitted transformer TabPFN, and the inherently interpretable Explainable Boosting Machine (EBM). Each configuration was assessed over 50 independent random train/test splits (70%/30%) under a strict leakage-free protocol in which all preprocessing, hyperparameter selection, and early stopping relied exclusively on inner validation subsets of the training partitions; all metrics are reported as mean ± ± SD.
Results: Under the leakage-free protocol, a group of four methods reached a statistically indistinguishable accuracy ceiling: on the cervical cancer dataset, ElasticNet (95.06% ± ± 4.66), EBM (94.94% ± ± 4.60), TabPFN (94.71% ± ± 4.64), and Random Forest (94.35% ± ± 5.82); on the Alzheimer's disease dataset, TabPFN (70.86% ± ± 10.26), EBM (70.67% ± ± 12.67), and ElasticNet (70.48% ± ± 9.76) led. TabNet reached 85.18% ± ± 11.10 (cervical) and 64.38% ± ± 11.23 (Alzheimer's) at its respective best levels -- below the top tier, but with the distinctive property of producing sparse, instance-level attention explanations as part of its forward computation. Supplementary analyses showed that self-supervised pre-training conferred no significant benefit at these cohort sizes, that results were robust to the choice between supervised differential-expression and unsupervised variance feature ranking, and that instance-level attention patterns varied substantially across splits, supporting their use as illustrative rather than confirmatory evidence.
Conclusions: In rigorously leakage-controlled small-cohort miRNA classification, strongly regularised classical models, glass-box additive models, and prior-fitted transformers converge to the same accuracy ceiling, and no evaluated method -- including attention-based deep learning -- surpasses it. Within this landscape, TabNet remains distinctive as the only evaluated deep architecture whose sequential attention mechanism produces sparse instance-level miRNA attributions as an inherent part of prediction, offering a trade-off between embedded interpretability and a moderate accuracy cost that we quantify precisely. These results argue for protocol-first evaluation and calibrated claims when deep tabular models are applied to small transcriptomic cohorts.
Related
- Interpretable Deep Learning of Endogenous Fluorophore Concentration and Entropy for Lung Tissue Classification
- Dysregulation of Expression of Nicotinic Acetylcholine Receptors and Ly6/uPAR Proteins in the Cingulate Gyrus of the Cerebral Cortex of Patients with Alzheimer's Disease.
- CPRISP: an interpretable deep learning framework predicting circRNA-RBP interactions and guiding therapeutic discovery in neurodegeneration
- Immune-related molecular links between asthma and Alzheimer’s disease identified by multi-omics, machine learning and single-cell analysis
- LitGene: an interpretable transformer model integrating text and ontology for gene representation learning
BibTeX
@article{haznedar2026attentive,
title = {{Attentive interpretable tabular learning (TabNet) for miRNA expression classification in cervical cancer and Alzheimer's disease}},
author = {Hilal Haznedar and Gözde Özsert Yiğit and Bülent Haznedar},
journal = {BMC Bioinformatics},
year = {2026},
month = oct,
doi = {10.1186/s12859-026-06706-7},
url = {https://doi.org/10.1186/s12859-026-06706-7},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Attentive interpretable tabular learning (TabNet) for miRNA expression classification in cervical cancer and Alzheimer's disease
AU - Hilal Haznedar
AU - Gözde Özsert Yiğit
AU - Bülent Haznedar
PY - 2026
DA - 2026/10/27
JO - BMC Bioinformatics
DO - 10.1186/s12859-026-06706-7
UR - https://doi.org/10.1186/s12859-026-06706-7
AB - Background and Objective: The classification of RNA sequencing (RNA-seq) data presents a fundamental challenge in computational biomedicine: raw read counts are high-dimensional, over-dispersed, and heteroscedastic, requiring careful preprocessing before machine learning models can be applied. Existing deep learning approaches for RNA-seq classification have largely overlooked the tabular nature of genomic expression data and have offered limited embedded interpretability. This study proposes TabNet, an attentive interpretable tabular learning architecture, as a framework for miRNA-based disease classification and evaluates its performance and interpretability against three established baseline models. Methods: Two publicly available miRNA expression datasets were examined: a cervical cancer cohort (58 samples, 714 miRNAs) and an Alzheimer's disease cohort (70 samples, 503 miRNAs). Raw count data were preprocessed following the MLSeq framework, encompassing DESeq median-ratio normalisation, near-zero variance filtering, supervised gene ranking based on DESeq differential expression analysis, and the DESeq variance-stabilising transformation (vst). Gene subsets corresponding to 5%, 10%, and 30% of the post-filtering feature space were evaluated. TabNet was benchmarked against eight baselines spanning four learning paradigms: a 1D Convolutional Neural Network (CNN), a Long Short-Term Memory network (LSTM), Extreme Gradient Boosting (XGBoost), Random Forest, ElasticNet-regularised logistic regression, a Multi-Layer Perceptron (MLP), the prior-fitted transformer TabPFN, and the inherently interpretable Explainable Boosting Machine (EBM). Each configuration was assessed over 50 independent random train/test splits (70%/30%) under a strict leakage-free protocol in which all preprocessing, hyperparameter selection, and early stopping relied exclusively on inner validation subsets of the training partitions; all metrics are reported as mean ± ± SD. Results: Under the leakage-free protocol, a group of four methods reached a statistically indistinguishable accuracy ceiling: on the cervical cancer dataset, ElasticNet (95.06% ± ± 4.66), EBM (94.94% ± ± 4.60), TabPFN (94.71% ± ± 4.64), and Random Forest (94.35% ± ± 5.82); on the Alzheimer's disease dataset, TabPFN (70.86% ± ± 10.26), EBM (70.67% ± ± 12.67), and ElasticNet (70.48% ± ± 9.76) led. TabNet reached 85.18% ± ± 11.10 (cervical) and 64.38% ± ± 11.23 (Alzheimer's) at its respective best levels -- below the top tier, but with the distinctive property of producing sparse, instance-level attention explanations as part of its forward computation. Supplementary analyses showed that self-supervised pre-training conferred no significant benefit at these cohort sizes, that results were robust to the choice between supervised differential-expression and unsupervised variance feature ranking, and that instance-level attention patterns varied substantially across splits, supporting their use as illustrative rather than confirmatory evidence. Conclusions: In rigorously leakage-controlled small-cohort miRNA classification, strongly regularised classical models, glass-box additive models, and prior-fitted transformers converge to the same accuracy ceiling, and no evaluated method -- including attention-based deep learning -- surpasses it. Within this landscape, TabNet remains distinctive as the only evaluated deep architecture whose sequential attention mechanism produces sparse instance-level miRNA attributions as an inherent part of prediction, offering a trade-off between embedded interpretability and a moderate accuracy cost that we quantify precisely. These results argue for protocol-first evaluation and calibrated claims when deep tabular models are applied to small transcriptomic cohorts.
SN - 1471-2105
ER -