Machine Learning for Anticancer Peptides
A machine learning model developed for determining anticancer activity of peptides
Patent-Pending Research Project
This project involves the development of a sophisticated machine learning approach to predict anticancer activity in antimicrobial peptides (AMPs). Using a random forest model based on amino acid and dipeptide features, I was able to screen 10,000 AMPs for potential anticancer properties.
Key Accomplishments
- Developed a random forest model that achieved 79.5% accuracy on the validation dataset
- Identified 1,648 peptides with promising anticancer activity
- Performed additional analyses including clustering, physicochemical property predictions, and 3D homology modeling
- Results suggest potential dual therapeutic effects of AMPs in cancer treatment
- Patent application filed for the prediction methodology
Technologies Used
- Machine learning (Random Forest algorithm)
- Bioinformatics tools for peptide sequence analysis
- Molecular modeling
- Statistical analysis
This research demonstrates the power of computational approaches in drug discovery, particularly for identifying dual-purpose peptides that may serve as both antimicrobial and anticancer agents.