Antibiotic resistance is one of emerging threats for the public health, which is projected to become the leading cause of deaths worldwide in coming decades. Discovery of new antibiotics is a slow and tedious process that requires screening mostly unculturable species for new mechanisms of defense from bacteria and it is laborious as you have to first produce the peptide, purify it and then test it one by one.
The goal of this master's project is to develop a machine learning model that can predict the antimicrobial properties of peptides based on their amino-acid sequence. Antimicrobial peptides are a promising class of substances for combating antibiotic resistance in bacteria, and their versatility and efficiency make them an attractive option. However, the vast number of possible peptide sequences makes discovery of new antimicrobial peptides difficult. Each peptide consists of amino-acids and there are 20 different amino-acids, so even a 10-amino-acid-long peptide has 10^20 variants. To address this challenge, we plan to use machine learning to analyze existing databases of antimicrobial peptide sequences and make predictions about the properties of new sequences.
The student working on this project will be responsible for:
1)Surveying the literature for available training datasets
2)Acquiring the datasets
3)Selecting the right predictive model and training it
4)Verifying the model with existing data
5) Making predictions for new peptide sequences that can be tested in the laboratory.
This project is a collaboration between the EEMCS and TNW faculties, and we are looking for an enthusiastic student to join us in this cutting-edge research.
Here is some reading about the computational methods in antimicrobial peptides: P. G. A. Aronica et al., “Computational Methods and Tools in Antimicrobial Peptide Research,” J. Chem. Inf. Model., vol. 61, no. 7, pp. 3172–3196, 2021, doi: 10.1021/acs.jcim.1c00175 ( but this paper contains a lot of non-ML methods to model the properties of peptides in general as well)
Contact person: