Computational Biology and Drug Design (CBDD)

Research in the CBDD group focuses on the development and application of computational methods to predict and analyse the modulation of protein and cell function by small organic molecules. Problems of interest include oncology biomarker discovery, modelling cancer pharmacogenomics, polypharmacology prediction and drug design (phenotypic, structure-based and ligand-based).

Biomarker Discovery

Targeted drugs, which inactivate specific molecular targets upon which cancer tumours rely to drive cell growth, have delivered therapies that are more specific and thus with generally less side-effects than traditional cytotoxic chemotherapy. Unfortunately, these therapies are only effective in some patients and our current ability to identify these responsive patients before administering the drug is still very limited. This differential drug response is not only due to heterogeneities between patients but also across tumour types.

The advent of Next Generation Sequencing (NGS) constitutes an unprecedented opportunity to study the molecular basis of this aspect of human variation. However, new computational methods are needed to predict the efficacy of a drug from the genomic and epigenomic status of patients and their tumours. Furthermore, these molecular profiles can also be used to predict prognosis given the clinical history of the patients from whom samples were taken. Therefore, our main goals in this area are: a) to investigate optimal methodologies to build predictive biomarkers of drug efficacy and prognosis and b) to apply these methodologies to leukemia, pancreatic and breast cancers in collaboration with preclinical and clinical scientists at the CRCM.



Modelling Cancer Pharmacogenomics

Following the recent availability of large-scale sensitivity data from screening about 120 drugs against a panel with more than 600 cancer cell lines, we introduced [1] a new type of computational models integrating genomic properties from the cell lines and chemical information from the drugs. This first Integrative Drug-Cell Sensitivity (IDCS) model can be used to provide useful estimations of the activity of a screened drug against a subset of these cell lines. Current research aims at improving the accuracy of this model in order to allow more challenging applications, such as obtaining accurate predictions against a particular cell line or predicting drug-gene associations for any research or approved drug.


Computational Drug Design

In addition to research intended to optimise the application of known drugs, there is a constant need to discover new drugs to treat cancer patients who are non-responsive, relapsed and/or have poor-prognosis. However, drug discovery is not possible without a way to identify molecules that modulate the biological function of a validated therapeutic target. There are now a range of computational methods that predict the biological activities of a molecule from ever-increasing volumes of relevant experimental data. When applied to Virtual Screening (VS), these computational methods can be used to search vast databases of candidate molecules for those likely to be active against the considered target. In practice, these tools have been able to discover drug leads and/or chemical probes in a wide range of targets and are particularly useful in those targets where High-Throughput Screening (HTS) performs poorly or it is not an option (e.g. technically not possible, too expensive or too slow). Other important applications of these in silico tools include the optimisation and polypharmacology prediction of drug leads.

Depending on the type of data available for the target, there are three classes of methods: cell-based, structure-based and ligand-based. Cell-based methods, such as those based on pharmacogenomics data, aim at identifying molecules inhibiting cancer cell growth [1]. Subsequently, one often wishes to find the protein targets of such phenotypic hits in order to understand their whole-cell activity. Beyond drug leads, this is also important to explain the efficacy and side-effects of development or even approved drugs. In order to address this challenge, we are currently investigating various methods for polypharmacology prediction exploiting recent chemogenomics databases.

If a structural model of the protein target is available (e.g. X-ray crystal structure), structure-based methods such as molecular docking can be used to predict whether and how the molecule binds the target. The latter is useful to design drug leads that modulate the molecular function of the target. The single most important limitation of docking is in the accuracy in predicting binding strength (re-scoring functions). In this area, we demonstrated [2] the advantages of machine-learning scoring functions over classical approaches. We also introduced [3] a user-friendly docking webserver implementing these advances. More recently, we revealed [4] that a more precise chemical description of the protein-ligand complex does not generally lead to enhanced binding prediction as it was previously thought. These novel methodologies have been successful [5] in discovering antibacterial inhibitors with new chemical scaffolds.

Finally, ligand-based methods relate the chemical structure of drug molecules with their bioactivities against a protein or cell target. In this area, we favour Ultrafast Shape Recognition (USR) molecular shape similarity [6] and their pharmacophoric extensions [7] because in practice these methods excel at finding bioactive molecules with new chemical scaffolds. For instance, USR found a high proportion of such inhibitors in an arylamine N-acetyltransferase target linked to breast cancer [8] and also in a phosphatase target linked to promoting metastasis in several types of cancer [9].

Drug Repositioning

Any of these computational methods can be applied to drug repositioning, which has become an area of intense interest in recent years. In drug repositioning, a drug is associated in some way to an indication other than that for which it was initially approved. This strategy has the advantage that all the toxicity-related requirements in the preclinical and clinical stages have already been met by the drug molecule. Therefore, if the predicted association is experimentally confirmed, the repositioned drug should follow a faster and cheaper route to Phase II clinical trials for the new indication. In a pilot study that used USR to screen the set of all FDA-approved drugs [10], only the best hit was experimentally tested for anti-cancer activity and this was enough to discover that the anti-hypertension drug Telmisartan inhibits the growth of colon cancer cell lines at the low micromolar level. Ongoing research includes the application of cell-based methods to unveil further drug repositioning opportunities as well as positioning a targeted drug candidate in responsive cancer subtypes.


[1] Menden, M., Iorio, F., Garnett, M., McDermott, U., Benes, C., Ballester, P.J. & Saez-Rodriguez, J. (2013) PLOS ONE 8: e61318.
[2] Ballester, P.J. & Mitchell, J.B.O. (2010) Bioinformatics 26, 1169-1175.
[3] Li H., Leung K.-S., Ballester P.J. & Wong M.-H. (2014) PLOS ONE 9: e85678.
[4] Ballester, P.J., Schreyer, A. & Blundell, T.L. (2014) Journal of Chemical Information and Modeling 54, 944-55.
[5] Ballester, P.J., Mangold, M., Howard, N.I., Marchese-Robinson, R.L., Abell, C., Blumberger, J. & Mitchell, J.B.O. (2012) Journal of the Royal Society Interface 9, 3196-207.
[6] Ballester, P.J. & Richards, W.G. (2007) Proceedings of the Royal Society A 463, 1307-1321
[7] Schreyer, A. & Blundell, T.L. (2012) Journal of Cheminformatics 4:27.
[8] Ballester, P.J., Westwood, I., Laurieri, N., Sim, E. & Richards, W.G. (2010) Journal of the Royal Society Interface 7, 335-342
[9] Hoeger, B., Diether, M., Ballester, P.J. & Köhn, M. (2014) European Journal of Medicinal Chemistry (In Press).
 [10] Patil, S.P., Ballester, P.J. & Kerezsi, C. (2014) Journal of Computer-Aided Molecular Design 28, 89-97.

The team

This team was created with the arrival of Dr Pedro Ballester to the CRCM in October 2014. Current members of the team are Dr Cuong Dang (postdoc), Dr Antonio Peon (postdoc), Ms Linh Nguyen (PhD student), Mr Nicolas Jaume (master student) and Dr Pedro Ballester (PI).

The team is interested in acting as a host for applications to CR2 permanent research positions from Inserm or CNRS as well as postdoctoral fellowships (e.g. EU Marie Curie, EMBO or HFSP programmes). Applications for PhD scholarships can also be supported. Prospective applicants must send a concise explanation of their research interests and a CV with publications to Pedro Ballester.

About the team leader

Pedro Ballester


Education and Work History

Awards and Honours

[last updated: July 2016]