The University of Manchester Featured PhD Programmes
Imperial College London Featured PhD Programmes
University of Cambridge Featured PhD Programmes

Deep Learning / Bioinformatics Approach for Protein-Protein Interaction Prediction


Faculty of Science, Engineering and Computing

About the Project

Since most molecular processes rely on protein–protein interactions (PPIs), knowledge of those interactions is extremely valuable for biomedical research and drug design. Despite the availability of high-throughput proteomics approaches, the human protein interactome is still largely incomplete. Therefore, ‘in-silico’ prediction (i.e. computer based) has become the only practical way of revealing the full extent of the human PPI network [1]. Although the development of bioinformatics methods allowing the prediction of such interactions is a very active field of investigation [2], existing approaches tend to focus on specific classes of interactions. For example, PPIs through β-sheet interfaces have been of particular interest [3], predominantly resulting from their potential to cause aggregation.

With the protein structure database, i.e. Protein Data Bank, approaching 200,000 entries – the majority of them revealing detailed information of PPIs -, sufficient data is now available to train the most data hungry machine learning approaches such as Deep Learning. The aim of this research is to exploit known 3D structure interactions in protein complexes to train a Deep Learning model able to predict if two proteins, defined only by their sequence, can form a dimer. Successful completion of the project requires addressing the following scientific objectives:
- Adaptation of existing Deep Learning architectures, such as convolutional neural network (CNN) and recurrent neural network (RNN), to design a suitable classification pipeline
- Implementation of the proposed Deep Learning based architecture on the Kingston University GPU farm
- PPI data extraction and model training
- Hyper-parameter tuning and classifier evaluation on standard PPI benchmark data sets

This project does not include funding. Applicants should have, at least, an Honours Degree at 2.1 or above (or equivalent) in Computer Science or related disciplines. In addition, they should have excellent programming skills in Matlab, Python, Java and/or C++. Fundamental knowledge of bioinformatics is not essential but desirable.

Qualified applicants are strongly encouraged to contact informally the supervising academic, Prof. Nebel (), to discuss the application. More on Prof. Nebel’s research group and activities can be found on his personal website: https://kunet.kingston.ac.uk/ku33185


Funding Notes

There is no funding for this project: applications can only be accepted from self-funded candidates

References

[1] Proteomics and Bioinformatics Soon to Resolve the Human Structural Interactome, J.-C. Nebel, Journal of Proteomics & Bioinformatics, 5(10): xi-xii, 2012
[2] Progress and Challenges in Predicting Protein Interfaces, R. Esmaielbeiki, K. Krawczyk, B. Knapp, J.-C. Nebel and C.M. Deane, Briefings in Bioinformatics, 17(1):117-131, 2016
[3] Coil conversion to β-strand induced by dimerisation, J. Laibe, A. Caffrey, M. Broutin, S. Guiglion,S.B. Pierscionek & J.-C. Nebel, PROTEINS: Structure, Function, and Bioinformatics, 86(12):1221-1230, 2018

Email Now

Insert previous message below for editing? 
You haven’t included a message. Providing a specific message means universities will take your enquiry more seriously and helps them provide the information you need.
Why not add a message here

The information you submit to Kingston University will only be used by them or their data partners to deal with your enquiry, according to their privacy notice. For more information on how we use and store your data, please read our privacy statement.

* required field

Your enquiry has been emailed successfully



Search Suggestions

Search Suggestions

Based on your current searches we recommend the following search filters.



FindAPhD. Copyright 2005-2020
All rights reserved.