ChEMBL

Here are details of some of the PhD studentship project ideas for the ChEMBL group. If you are interested in studying in any of these areas at the EMBL-EBI outstation in Hinxton, please contact us. The underlying theme of the projects is to provide infrastructure components built upon the ChEMBL databases for applied and translational drug discovery. They all offer the opportunity to study and develop cutting-edge Open Source drug discovery technologies in a highly collaborative, diverse and international setting.

1. Non-human Secreted Proteins as New Therapies.

Surprisingly, several classes of very important human biological therapeutics are derived from non-human sources; for example, hirudin from medicinal leaches. This project will apply data-mining approaches to characterise the features for immunotolerance on non-host proteins, and then develop a series of approaches to identify potential candidate therapeutic proteins from a wide variety of organisms (e.g. Mammals, Ticks, Flukes, Nematodes, etc.). The project will integrate sequence and structure-based data mining methods along with KDD/data-mining methods to predict function, pharmacokinetic and affinity properties.

2. Monoclonal Antibody Drug ‘Rescue’

Monoclonal antibody drugs (mAbs) are viewed as highly specific therapies, with low clinical failure rates. However, attrition for mAb therapies is often expensive and occurs at a late and expensive stage in their clinical progression, with success often crucially dependent on choice of disease model and trial design. As part of our CandiStore project, we have accumulated a unique set of clinical stage mAbs, and their targets ligands. This project will apply modelling, docking, text-mining and KDD approaches to understand and predict new clinical applications of previously failed mAbs. The project will also attempt to discover general features for success/failure of this important class of therapy.

3. Automated Drug Design (‘Robot-Chemist’)

The discovery and optimisation of novel, well tolerated drugs is becoming an increasingly important commercial challenge. We have assembled a large training set of SAR data in our StARlite database, and have produced proof-of-principle applications in areas such as bioisostere discovery, affinity optimisation, etc. This project will build a catalog of empirically observed and synthetically tractable transformations from a given chemical starting point, and attempt to objectively score their likely effect on bioactivity, in particular drug like properties such as metabolism, frank toxicity, absorption, etc. This project will provide an excellent introduction to the principles of medicinal chemistry from a compound design perspective, and also a firm grounding in a broad variety of KDD approaches.

4. Drug Design Strategies for Robustness to Acquired Resistance.

Acquired resistance, the selection of mutant forms of a target under the selective pressure of a cidal drug, is of increasing importance in both anti-microbial and anti-cancer targeted therapies. In the vast majority of cases, this resistance can be understood at a structural level, once the 3-D structure of the drug-target complex is established. This project will provide an integrative approach for the prediction of alternate functional forms of a target, under simulated evolutionary pressure of drug binding. Techniques such as comparative modelling, sequence analysis, Monte-Carlo simulation and QSAR approaches will be applied, and the methods then used at genome scale to identify targets and compound design strategies likely to be robust to acquired resistance. Depending on progress made during the major part of the project, the methods could be applied to understand differential drug response caused by genetic diversity/cSNPs in the human genome for currently approved therapies.