Peptide Resources

SPdb is a professional database focusing on signal peptides, integrating experimentally validated and computationally predicted signal peptide sequences. Data originates from Swiss-Prot (now part of UniProt) and EMBL nucleotide databases, updated semi-automatically with manual verification for accuracy. The latest release 3.2 contains 18,146 entries, including 2,584 experimentally validated signal sequences and 15,562 entries that fail to meet filtering criteria or contain unverified sequences. Supporting signal peptide function analysis and protein localization prediction, it provides data support for research on protein targeting mechanisms in prokaryotic and eukaryotic cells.

Stack-AAgP is an ensemble-learning-based model for anti-angiogenic peptide (AAPs) prediction, constructing 24 baseline models via six machine learning algorithms (Random Forest, XGBoost, etc.) and four feature encodings (PAAC, APAAC, etc.), then training meta-classifiers with their predicted probabilities.

Stack-AAgP is an ensemble-learning-based model for anti-angiogenic peptide (AAPs) prediction, constructing 24 baseline models via six machine learning algorithms (Random Forest, XGBoost, etc.) and four feature encodings (PAAC, APAAC, etc.), then training meta-classifiers with their predicted probabilities.

Stack-ACPred is a novel predictor for accurate identification of anticancer peptides (ACPs), employing three feature encoding strategies: Segmented Position-Specific Scoring Matrix (SegPSSM), Pseudo Amino Acid Composition (PsePSSM), and Extended PseAAC, fusing evolutionary profiles and physicochemical information.

StackCPPred is a computational tool for predicting cell-penetrating peptides and their uptake efficiency, leveraging pairwise energy content-based features (RECM-composition, PseRECM, RECM-DWT) and stacking machine learning methods.