A 4-Qubit Variational Quantum Classifier for Sequence-Specific Classification of DNMT3A and DNMT3B Flanking Methylation Preferences

Murat Isik; Yusuf Moinuddin; Humaira Gowher; Sabre Kais

cs.LGOct 12, 2026

A 4-Qubit Variational Quantum Classifier for Sequence-Specific Classification of DNMT3A and DNMT3B Flanking Methylation Preferences

Authors: Murat Isik, Yusuf Moinuddin, Humaira Gowher, Sabre Kais

Organizations: Purdue University, West Lafayette, USA · NC State University, Raleigh, USA

Abstract

DNA methylation is a critical epigenetic modification that regulates gene expression and genome stability, primarily catalyzed by the DNA methyltransferase 3 (DNMT3) family of enzymes. Distinguishing precise sequence-specific flanking preferences between DNMT3A and DNMT3B remains computationally challenging due to the complex and high-dimensional nature of biochemical sequence data. In this study, we propose a novel Quantum Machine Learning (QML) approach to classify DNA methylation preferences, leveraging a Variational Quantum Classifier (VQC). Using experimentally derived deep enzymology data, we constructed a dataset containing 4,096 unique NNNCGNNN motifs. We evaluated a four-qubit VQC using a custom NumPy exact-statevector simulator. The described two-layer circuit contains 24 trainable rotations, eight CNOT gates, and eight input-encoding rotations, totaling 40 gates before measurement. Across five initialization seeds, the revised VQC evaluation achieved a ROC-AUC of 0.661 pm 0.009, where variability denotes sample standard deviation. The held-out test partition is approximately balanced, containing 200 DNMT3A-labeled motifs and 210 DNMT3B-labeled motifs. An always-DNMT3B predictor achieves 51.2% accuracy and a balanced accuracy of 0.500 on this test partition. This proof-of-concept establishes a foundation for utilizing near-term quantum architectures in complex epigenetic classification tasks, paving the way for hybrid quantum-classical pipelines in bioinformatics.

Figures & tables

Fig. 1: Biological context of DNMT3A and DNMT3B methylation preferences. The schematic details the target 8-mer motif ( $NNNCGNNN$ ) alongside the functional differentiation within the DNMT family, contrasting de novo methylation by DNMT3A and DNMT3B with maintenance methylation by DNMT1. Methyl group addition to cytosine at CpG sites produces 5-methylcytosine (5mC).

Fig. 2: Dataset preparation and evaluation workflow for DNMT3A/DNMT3B preference classification. The revised dataset contains 4,096 unique $NNNCGNNN$ motifs, with one entry per motif. A fixed stratified split assigns 3,276 motifs to training, 410 to validation, and 410 to testing, with no motif overlap between partitions. Validation data are used for model and checkpoint selection; test data are reserved for final evaluation.

Fig. 3: Hybrid Quantum-Classical Variational Classifier Architecture. The diagram illustrates the complete forward data pass from an 8-mer genomic motif input through nucleotide-pair condensation and ordinal encoding (A $\mapsto$ 0, C $\mapsto$ 1, G $\mapsto$ 2, T $\mapsto$ 3), 4D feature mapping to $x\in[-\pi,\pi]^{4}$ , 2-layer parameterized quantum circuit execution, 4-qubit Pauli-Z measurement, linear readout, sigmoid activation, and weighted binary cross-entropy backpropagation. The sigmoid output represents the DNMT3B class score: $p\geq 0.5$ predicts DNMT3B (label 1), while $p<0.5$ predicts DNMT3A (label 0).

Fig. 4: Earlier evaluation plots, retained separately from the revised analysis. (a) ROC curve. (b) Precision–recall curve. (c) Multi-seed performance distributions. (d) Changes in ROC-AUC under feature masking. Embedded values have not been reconciled with the revised test partition.

Hardware/Circuit Parameter	Observed Metric Value
Register Allocation	4 Operational Qubits
Evaluation Resolution	256 Shots
Total Structural Gate Count	36 Gates
Parameterized Rotation Gates	24 ( ry : 8, rx : 8, rz : 8)
Entangling Operations	8 CNOT ( cx ) Gates
Measurement Operators	4 Pauli-Z ( $\langle\mathcal{Z}_{i}\rangle$ ) Operators

TABLE I: Earlier reported circuit specifications and Qiskit Aer profiling measurements.

Seed	Best Epoch	Accuracy	Bal. Acc.	ROC-AUC	PR-AUC	F1 (DNMT3B)
0	25	0.671	0.584	0.648	0.269	0.306
1	15	0.662	0.652	0.666	0.322	0.375
2	13	0.729	0.655	0.657	0.310	0.391
3	14	0.715	0.609	0.664	0.281	0.337
4	7	0.744	0.627	0.652	0.293	0.361
Mean $\pm$ Std		$0.704\pm.032$	$0.625\pm.027$	$0.657\pm.007$	$0.295\pm.019$	$0.354\pm.030$

TABLE II: Earlier per-seed evaluation results, retained separately from the revised analysis.

Quantum Machine Learning

Quantum machine learning uses quantum circuits and quantum data to build learning algorithms and study their computational and statistical limits.

See all 256 papers

MEDNA-DFM: A Dual-View FiLM-MoE Model for Explainable DNA Methylation Prediction
Feb 26, 2026Yi He, Yina Cao, Jixiu Zhai +3Genomics Sequence Modeling
Cuiying Honors College, Lanzhou University, Lanzhou, Gansu, China · School of Mathematics and Statistics, Lanzhou University, Lanzhou, Gansu, China · Cuiying Honors College, Lanzhou University, Lanzhou, Gansu, China. +7
A Boltzmann-machine-enhanced Transformer For DNA Sequence Classification
Mar 27, 2026Zhixuan Cao, Yishu Xu, Xuang WUTransformer Transformer Attention
Tsinghua University · University of California, Berkeley
A Specialized Importance-Aware Quantum Convolutional Neural Network with Ring-Topology (IA-QCNN) for MGMT Promoter Methylation Prediction in Glioblastoma
Apr 24, 2026Emine Akpinar, Murat OduncuogluQuantum Convolutional Neural Networks Hybrid Quantum-Classical ML
Department of Physics, Yildiz Technical University, Istanbul, Turkey

Feb 26, 2026J/K move · Enter open · S save

MEDNA-DFM: A Dual-View FiLM-MoE Model for Explainable DNA Methylation Prediction

Yi He, Yina Cao, Jixiu Zhai, Di Wang+2

Cuiying Honors College, Lanzhou University, Lanzhou, Gansu, China · School of Mathematics and Statistics, Lanzhou University, Lanzhou, Gansu, China · Cuiying Honors College, Lanzhou University, Lanzhou, Gansu, China.+7

Genomics Sequence Modeling

Open paper AskPDF arXiv

Accurate computational identification of DNA methylation is essential for understanding epigenetic regulation. Although deep learning excels in this binary classification task, its "black-box" nature impedes biological insight. We address this by introducing a high-performance model MEDNA-DFM, alongside mechanism-inspired signal purification algorithms. Our investigation demonstrates that MEDNA-DFM effectively captures conserved methylation patterns, achieving robust distinction across diverse species. Validation on external independent datasets confirms that the model's generalization is driven by conserved intrinsic motifs (e.g., GC content) rather than phylogenetic proximity. Furthermore, applying our developed algorithms extracted motifs with significantly higher reliability than prior studies. Finally, empirical evidence from a Drosophila 6mA case study prompted us to propose a "sequence-structure synergy" hypothesis, suggesting that the GAGG core motif and an upstream A-tract element function cooperatively. We further validated this hypothesis via in silico mutagenesis, confirming that the ablation of either or both elements significantly degrades the model's recognition capabilities. This work provides a powerful tool for methylation prediction and demonstrates how explainable deep learning can drive both methodological innovation and the generation of biological hypotheses.

Abstract

Quantum Machine Learning

Explore similar work

MEDNA-DFM: A Dual-View FiLM-MoE Model for Explainable DNA Methylation Prediction

A Boltzmann-machine-enhanced Transformer For DNA Sequence Classification

A Specialized Importance-Aware Quantum Convolutional Neural Network with Ring-Topology (IA-QCNN) for MGMT Promoter Methylation Prediction in Glioblastoma