Researchers from the University of the Philippines Diliman College of Science have developed an artificial intelligence tool that could significantly speed up the discovery of new antibiotics. The innovation focuses on antibacterial peptides, small proteins that can potentially be used to fight harmful bacteria.
The tool, called ISCAPE, was created by Remmer Salas, Portia Mahal Sabido, and Ricky Nellas from the university’s Institute of Chemistry. It is designed to predict whether a peptide can kill or inhibit Escherichia coli, a bacterium responsible for illnesses such as food poisoning, diarrhea, and urinary tract infections.
Cutting Down Trial and Error
Developing new antibacterial compounds has traditionally been a long and resource-intensive process, requiring scientists to synthesize and test candidates individually. ISCAPE offers a faster alternative by evaluating molecules using a Simplified Molecular Input Line Entry System (SMILES) string.
Through machine learning, the system analyzes existing data to identify patterns that distinguish effective peptides from inactive ones. It can also highlight specific molecular features that contribute to antibacterial activity. This allows researchers to design better candidates early on while reducing the need for repeated trial-and-error experiments.
Despite its capabilities, the developers emphasize that ISCAPE is not meant to replace laboratory testing. Instead, it serves as a guide, helping scientists prioritize the most promising compounds for further study.
Potential Beyond One Bacterium
The researchers say the tool could play a role in addressing antimicrobial resistance by improving early-stage screening of potential antibiotics. While ISCAPE is currently focused on E. coli, it can be adapted to other bacteria or bioactive peptides if trained with high-quality, experimentally validated datasets.
ISCAPE is publicly available through Hugging Face Spaces, with its training data and large-scale prediction code accessible on GitHub. The full study has been published in the Journal of Molecular Graphics and Modelling, highlighting its potential to support faster and more targeted antibiotic development.
