Peptide Resources

AI4AVP is a deep learning-based platform for antiviral peptide (AVPs) prediction, leveraging generative adversarial networks (GANs) and convolutional neural networks (CNNs) for efficient peptide sequence activity identification. The model innovatively uses GANs to expand positive training data, addressing the scarcity of experimental AVP data, while CNNs automatically extract spatial feature patterns (e.g., hydrophobic cluster distribution, charge periodicity) from sequences. 

AIE-ECM is a predictive tool for accurate identification of anti-inflammatory peptides (AIEs), encoding peptide sequences based on three feature representations: amino acid contact, position-specific scoring matrix (PSSM), and physicochemical properties. It constructs base classifiers via multi-feature extraction models and feature selection, then learns discriminative feature combinations through an Ensemble Classifier Chain model. 

An anti-inflammatory peptide (AIPs) prediction framework integrating ensemble machine learning and deep learning, constructing three individual models: Extremely Randomized Trees (ET), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN) with attention mechanism, then fusing results via stacking. Integrating multiple sequence encodings, it leverages ET for nonlinear feature relationships, GRU for sequential dependencies, and CNN with attention for local key feature extraction, achieving 0.757 accuracy, 0.500 MCC, and 0.707 F1-score on independent test sets—outperforming existing methods. Its feature interpretation mechanism provides explainable support for AIPs design, facilitating anti-inflammatory therapy development.

AIPpred is a random forest (RF)-based tool for predicting anti-inflammatory peptides (AIPs), trained with 354 optimal features. The model systematically studied the contribution of individual composition features [amino acid composition, dipeptide composition (DPC), amino acid index, chain-transition-distribution, and physicochemical properties] in AIP prediction.

AIPPT is an intelligent and computationally efficient predictor for reliable identification of anti-inflammatory peptides (AIP), introducing a novel stacking framework that integrates four feature encodings. Feature importance is evaluated via LightGBM to construct an optimal subset, fed into three base classifiers (e.g., Random Forest, SVM), whose output probabilities are integrated by a meta-classifier (e.g., Logistic Regression) to form a two-layer stacking model for precise AIP prediction.