Drug Protein Interaction - MA - Kuster Lab
Title: Drug Target Prediction
Type: MA
Category: ML,DL
Programming language: python (Tensorflow/pytorch)
Language: [ English ]
Prior experience: [ Experience With Python, Experience with DL framework (Tensorflow/Pytorch), Chemical/Biological background is not required ]
Complexity/Risk: high
Contact person: Wassim Gabriel
Brief background description (couple of sentences + literature):
Discovery of potential drugs is an expensive and time consuming process. For decades, researchers and industry have tried to develop methods to predict which drug is going to interact with which protein(s) - also referred to as targets in this context. However, the complex nature of the interactions are very difficult to model. Deep learning could offer an alternative route as it does not require heavy feature engineering and modeling. The development of new methods to predict the interactions between drugs and targets has the potential to significantly increase the throughput of drug discovery platforms while at the same time lower overall cost.
Literature:
Brief description of the project (couple of sentences):
The goal of this project is to further extend or develop new deep learning models that learn and are able to predict the interaction between the proteins and drugs. Two routes are possible, either the prediction of whether a drug will bind to a protein (classification) or the prediction of their binding affinity (regression). For learning, we are able to make use of a large dataset we have generated in house. (Note: Because it’s deep learning, there is no need to have a solid background in chemistry/biology to be able to work on this project.)
Expected result:
Deep Learning model given protein and drug information as an input that can predict i.e. whether binding occurs or at best the affinity value. The results could lead to a short article in a peer-reviewed journal.