ATP-Pred: Prediction of Protein-ATP Binding Residues via Fusion of Residue-Level Embeddings and Kolmogorov–Arnold Network

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: March 22, 2025

Accurately identifying protein-ATP binding residues is essential for understanding biological processes and designing drugs. However, current sequence-based methods have limitations, such as difficulties in extracting discriminative features the need more efficient algorithms. Additionally, based on multiple sequence alignments often face challenges handling large-scale predictions. To address these issues, we developed ATP-Pred, a method predicting ATP-binding proteins. This model applies transfer learning by using two recently pretrain protein language models, Ankh ProstT5, to extract residue-level embeddings that capture functionality. ATP-Pred also integrates CNN-BiLSTM network Kolmogorov–Arnold build prediction model. handle data imbalance, introduced weighted focal loss function. Experimental results three independent test sets showed outperforms most existing methods. Its generalizability was further validated four protein-mononucleotide residue sets, where it delivered promising results. These findings suggest robust reliable predictor.

Language: Английский

---

CasPro-ESM2: Accurate identification of Cas proteins integrating pre-trained protein language model and multi-scale convolutional neural network

International Journal of Biological Macromolecules, Journal Year: 2025, Volume and Issue: unknown, P. 142309 - 142309

Published: March 1, 2025

Language: Английский

---

PLPTP: A Motif-Based Interpretable Deep Learning Framework Based on Protein Language Models for Peptide Toxicity Prediction

Journal of Molecular Biology, Journal Year: 2025, Volume and Issue: unknown, P. 169115 - 169115

Published: March 1, 2025

Language: Английский

---

PLM-IL4: Enhancing IL-4-inducing Peptide Prediction with Protein Language Model

Computational Biology and Chemistry, Journal Year: 2025, Volume and Issue: unknown, P. 108448 - 108448

Published: April 1, 2025

Language: Английский

---

NeuroPred-AIMP: Multimodal Deep Learning for Neuropeptide Prediction via Protein Language Modeling and Temporal Convolutional Networks

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: April 21, 2025

Neuropeptides are key signaling molecules that regulate fundamental physiological processes ranging from metabolism to cognitive function. However, accurate identification is a huge challenge due sequence heterogeneity, obscured functional motifs and limited experimentally validated data. Accurate of neuropeptides critical for advancing neurological disease therapeutics peptide-based drug design. Existing neuropeptide methods rely on manual features combined with traditional machine learning methods, which difficult capture the deep patterns sequences. To address these limitations, we propose NeuroPred-AIMP (adaptive integrated multimodal predictor), an interpretable model synergizes global semantic representation protein language (ESM) multiscale structural temporal convolutional network (TCN). The introduced adaptive fusion mechanism residual enhancement dynamically recalibrate feature contributions, achieve robust integration evolutionary local information. experimental results demonstrated proposed showed excellent comprehensive performance independence test set, accuracy 92.3% AUROC 0.974. Simultaneously, good balance in ability identify positive negative samples, sensitivity 92.6% specificity 92.1%, difference less than 0.5%. result fully confirms effectiveness strategy task recognition.

Language: Английский

---

Transformer-based deep learning enables improved B-cell epitope prediction in parasitic pathogens: A proof-of-concept study on Fasciola hepatica

PLoS neglected tropical diseases, Journal Year: 2025, Volume and Issue: 19(4), P. e0012985 - e0012985

Published: April 29, 2025

Background The identification of B-cell epitopes (BCEs) is fundamental to advancing epitope-based vaccine design, therapeutic antibody development, and diagnostics, such as in neglected tropical diseases caused by parasitic pathogens. However, the structural complexity parasite antigens high cost experimental validation present certain challenges. Advances Artificial Intelligence (AI)-driven protein engineering, particularly through machine learning deep learning, offer efficient solutions enhance prediction accuracy reduce costs. Methodology/Principal findings Here, we deepBCE-Parasite, a Transformer-based model designed predict linear BCEs from peptide sequences. By leveraging state-of-the-art self-attention mechanism, achieved remarkable predictive performance, achieving an approximately 81% AUC 0.90 both 10-fold cross-validation independent testing. Comparative analyses against 12 handcrafted features four conventional algorithms (GNB, SVM, RF, LGBM) highlighted superior power model. As case study, deepBCE-Parasite predicted eight leucine aminopeptidase (LAP) Fasciola hepatica proteomic data. Dot-blot immunoassays confirmed specific binding seven synthetic peptides positive sera, validating their IgG reactivity demonstrating model’s efficacy BCE prediction. Conclusions/Significance demonstrates excellent performance predicting across diverse pathogens, offering valuable tool for design vaccines, antibodies, diagnostic applications parasitology.

Language: Английский

---

Voting-ac4C:Pre-trained large RNA language model enhances RNA N4-acetylcytidine site prediction

International Journal of Biological Macromolecules, Journal Year: 2024, Volume and Issue: 282, P. 136940 - 136940

Published: Oct. 30, 2024

Language: Английский

---

ToxGIN: an In silico prediction model for peptide toxicity via graph isomorphism networks integrating peptide sequence and structure information

Briefings in Bioinformatics, Journal Year: 2024, Volume and Issue: 25(6)

Published: Sept. 23, 2024

Abstract Peptide drugs have demonstrated enormous potential in treating a variety of diseases, yet toxicity prediction remains significant challenge drug development. Existing models for peptide largely rely on sequence information and often neglect the three-dimensional (3D) structures peptides. This study introduced novel model short prediction, named ToxGIN. The utilizes Graph Isomorphism Network (GIN), integrating underlying amino acid composition 3D ToxGIN comprises three primary modules: (i) Sequence processing module, converting sequences into nodes edges; (ii) Feature extraction utilizing GIN to learn discriminative features from (iii) Classification employing fully connected classifier prediction. performed well independent test set with F1 score = 0.83, AUROC 0.91, Matthews correlation coefficient 0.68, better than existing toxicity. These results validated effectiveness structural data using proposed can be freely accessible at https://github.com/cihebiyql/ToxGIN.

Language: Английский

---

SenSeqNet: A Deep Learning Framework for Cellular Senescence Detection from Protein Sequences

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 1, 2024

Abstract Cellular senescence, characterized by the irreversible cessation of division in normally proliferating cells due to various stressors, presents a significant challenge treatment age-related diseases. Understanding and accurately detecting cellular senescence is crucial for identifying potential therapeutic targets. However, traditional wet lab assays are time-consuming labor-intensive, limiting research drug development efficiency. There an urgent need computational tools allowing swift accurate detection from protein sequences. We propose SenSeqNet, novel deep learning framework directly The begins with feature extraction using Evolutionarily Scaled Model (ESM-2), state-of-the-art language model that captures evolutionary information complex sequence patterns. extracted embeddings then passed through hybrid architecture consisting long short-term memory (LSTM) networks convolutional neural (CNNs) further refine learn embedded information. SenSeqNet achieved final accuracy 83.55% on independent testing, surpassing machine architectures. This performance underscoring robustness effectiveness These results provide solid foundation future aging therapeutics.

Language: Английский

---