Peptide Resources

iAIPs is a random forest-based model for identifying anti-inflammatory peptides (AIPs), achieving precise prediction via a three-step strategy: firstly, encoding original samples using three feature extraction methods—g-gap dipeptide composition (GDC), dipeptide deviation from expected mean (DDE), and amino acid composition (AAC); secondly, ranking features via analysis of variance (ANOVA) and generating optimal subsets through incremental feature selection; finally, inputting selected features into random forest classifiers.

iAMP-Attenpred is the first AMP predictor to employ BERT model from natural language processing for feature encoding, combining 1D CNN, BiLSTM, and attention mechanism to build a composite model, achieving efficient AMP classification through BERT-based feature extraction and multi-model fusion, outperforming existing predictors.

iAMPCN is a convolutional neural network (CNN)-based framework for identifying antimicrobial peptides (AMPs) and their functional activities, designed to address the limitation of existing methods in predicting multi-dimensional AMP functions. Integrating four sequence features (amino acid composition, physicochemical properties, sequence motifs, evolutionary information), the model constructs a multi-class prediction system covering 22 functional activities (e.g., antiviral, antifungal, anticancer).

IAMPE is a web-based platform for predicting antimicrobial peptides by clustering amino acids via 13CNMR spectroscopy to build feature vectors (composition, transition, distribution) combined with physicochemical properties, using NB, KNN, SVM and other ML algorithms, sourced from CAMP, LAMP databases, enhancing prediction performance through integrated features.

IF-AIP is an anti-inflammatory peptide (AIPs) identification model based on voting classifiers, integrating eight feature encodings (covering amino acid composition, physicochemical properties, etc.) to construct hybrid feature sets, combining five traditional machine learning classifiers (Random Forest, SVM, etc.) for voting integration, and optimizing the final model via feature selection algorithms.