EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

VIRGO-Motif: A Hybrid Stacked Ensemble and Reinforcement Learning Framework for Motif Discovery

M. Shilpa, C. Nandini

Engineering, Technology & Applied Science Research · 11.08.2026 · 16 (4) · 38092-38103

DOI: 10.48084/etasr.19292 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Engineering, Technology & Applied Science Research, online erschienen am 11.08.2026. Verfasst von M. Shilpa und 1 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The identification of Transcription Factor Binding Sites (TFBSs), i.e., motifs, is known as the motif discovery problem. Although deep learning algorithms have achieved promising performance in motif discovery, there is still scope for improving prediction accuracy. In this research paper, we propose Vision Transformer-driven Intelligent Reinforcement-guided Genetic Optimization (VIRGO)-Motif, a hybrid framework that integrates a stacked ensemble of deep learning models with Reinforcement Learning (RL) and Genetic Algorithms (GAs) for accurate and biologically valid DNA motif discovery. In the proposed framework, three diverse models, i.e., Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and Vision Transformer (ViT), are stacked. The meta learner, logistic regression, is trained on Out-of-Fold (OOF) predictions generated through stratified K-fold cross-validation. The candidate motifs are then optimized using an RL strategy combined with a metaheuristic GA. The proposed model was trained and tested using a benchmark set of ENCODE ChIP-seq datasets. Experimental results demonstrated an accuracy of 87.10%, a Precision–Recall Area Under the Curve (PR-AUC) of 94.05%, and a Receiver Operating Characteristic Area Under the Curve (ROC-AUC) of 94.09%, representing a significant performance improvement over state-of-the-art deep learning models such as DeepBind and DeepSTF. The biological significance of the discovered motifs was validated by comparing them with known motifs in the JASPAR database using the TOMTOM tool.

Zitieren

APA

Shilpa, M., & Nandini, C. (2026). VIRGO-Motif: A Hybrid Stacked Ensemble and Reinforcement Learning Framework for Motif Discovery. Engineering, Technology & Applied Science Research, 16(4), 38092-38103. https://doi.org/10.48084/etasr.19292

Vancouver

Shilpa M, Nandini C. VIRGO-Motif: A Hybrid Stacked Ensemble and Reinforcement Learning Framework for Motif Discovery. Engineering, Technology & Applied Science Research. 2026;16(4):38092-38103. doi:10.48084/etasr.19292

BibTeX
@article{Shilpa2026,
  author = {Shilpa, M. and Nandini, C.},
  title = {VIRGO-Motif: A Hybrid Stacked Ensemble and Reinforcement Learning Framework for Motif Discovery},
  journal = {Engineering, Technology & Applied Science Research},
  year = {2026},
  volume = {16},
  number = {4},
  pages = {38092--38103},
  doi = {10.48084/etasr.19292}
}
Kostenlos starten →