A study presents an in silico method for predicting the toxicity of peptides and proteins, called ToxinPred. The method uses machine learning and various features such as amino acid composition, dipeptide composition, and motifs to predict whether a peptide is toxic or not. The study collected toxic peptides from multiple databases and non-toxic peptides from SwissProt and TrEMBL. They developed models using support vector machines (SVM) and found that the dipeptide-based model achieved an accuracy of 94.50% with a Matthews correlation coefficient (MCC) of 0.88. They also developed a hybrid model combining motif information with SVM results, which achieved an accuracy of 98.41% with an MCC of 0.96. The study also developed a web server, ToxinPred, which allows users to predict the toxicity of peptides, design peptides with desired toxicity, and identify toxic regions in proteins. The results show that ToxinPred is a reliable tool for predicting the toxicity of peptides and proteins, which can aid in the development of safer therapeutic peptides and the discovery of toxic regions in proteins. The study highlights the importance of computational methods in the development of peptide-based drugs and the need for tools that can predict toxicity to improve the safety and efficacy of these drugs.A study presents an in silico method for predicting the toxicity of peptides and proteins, called ToxinPred. The method uses machine learning and various features such as amino acid composition, dipeptide composition, and motifs to predict whether a peptide is toxic or not. The study collected toxic peptides from multiple databases and non-toxic peptides from SwissProt and TrEMBL. They developed models using support vector machines (SVM) and found that the dipeptide-based model achieved an accuracy of 94.50% with a Matthews correlation coefficient (MCC) of 0.88. They also developed a hybrid model combining motif information with SVM results, which achieved an accuracy of 98.41% with an MCC of 0.96. The study also developed a web server, ToxinPred, which allows users to predict the toxicity of peptides, design peptides with desired toxicity, and identify toxic regions in proteins. The results show that ToxinPred is a reliable tool for predicting the toxicity of peptides and proteins, which can aid in the development of safer therapeutic peptides and the discovery of toxic regions in proteins. The study highlights the importance of computational methods in the development of peptide-based drugs and the need for tools that can predict toxicity to improve the safety and efficacy of these drugs.