ToxinPred: In Silico Approach for Predicting Toxicity of Peptides and Proteins

ToxinPred: In Silico Approach for Predicting Toxicity of Peptides and Proteins ToxinPred is a computational tool developed to predict, design, and scan toxic peptides and proteins. The server helps identify whether a peptide is toxic or non-toxic, supports the design of peptide analogues with reduced or enhanced toxicity, and allows scanning of protein sequences to detect toxic regions. This resource is useful for peptide/protein-based drug discovery because toxicity is one of the major limitations in developing therapeutic peptides and proteins. Web Server: (https://webs.iiitd.edu.in/raghava/toxinpred/) Citation Gupta, S., Kapoor, P., Chaudhary, K., Gautam, A., Kumar, R., Open Source Drug Discovery Consortium, and Raghava, G. P. S. In Silico Approach for Predicting Toxicity of Peptides and Proteins. PLOS ONE, 8(9), e73957, 2013. https://doi.org/10.1371/journal.pone.0073957 About the Research Peptide and protein-based therapeutics have gained major importance because of their high specificity, biological activity, lower production cost, and better penetration compared to many small molecules. However, toxicity, immunogenicity, and stability remain major concerns in peptide-based drug development. ToxinPred was developed to predict the toxicity of peptides and proteins before experimental synthesis, helping researchers save time and cost during therapeutic peptide design. Data Compilation: Toxic peptides of 35 or fewer residues were collected from multiple toxin-related databases, including ATDB, DBETH, BTXpred, NTXpred, ArachnoServer, ConoServer, and Swiss-Prot. Non-toxic peptides were obtained from Swiss-Prot and TrEMBL. Methodology: ToxinPred uses machine learning-based models, mainly Support Vector Machine models, trained on peptide sequence features such as amino acid composition, dipeptide composition, binary profile patterns, toxic motifs, and quantitative matrices.