Computational Biology

The UPDDI prides itself on its computational expertise and its many collaborations with members of the Department of Computational and Systems Biology (CSB) faculty. Specific areas of excellence include:

Target assessment

Controlled modulation of molecular activity remains at the heart of drug discovery. Our experts employ a range of computational tools to understand the structure, dynamics and function of existing drug targets, and to explore the potential of novel targets. We develop and use coarse-grained models with simplified potentials to investigate global protein dynamics and to identify sites that may enable allosteric modulation of function. We use conventional molecular dynamics (MD) simulations to understand protein function at atomic resolution and use MD in the presence of small organic solvent molecules to discover binding pockets on protein surfaces. Our DCSB collaborators have developed accurate and efficient methods for virtually screening millions of compounds in seconds.

Systems Biology:

Systems biology, the holistic approach to understanding biology as emerging from a complex network of chemical interactions, is a central component of QSP. Our systems biology experts computationally infer relationships among proteins, predict phenotypic effects of small molecule perturbagens, and identify pathways of disease progression. We are actively developing new methods and software for incorporating diverse omics data into predictive models, and for understanding the origins of heterogeneity and disease. In parallel with the UPDDI’s experimental studies, we construct mechanistic pathway models to generate and guide the selection of hypotheses that are consistent with the data. Quantitative systems modeling thus provides us with previously unattainable insights into the nature of disease and drug activity.

Computational Pathology

Genetic variation is an indicator of tumor aggression and resistance to treatment, and phenotypic variation among cancer cells is also increasingly associated with poor prognosis. In collaboration with our imaging team and GE Global Research, our experts are developing computational algorithms, software and web tools for analyzing cellular heterogeneity in imaged tumors. Our state-of-the-art software includes methods for cell segmentation and classification, as well as for the analysis of spatial heterogeneity. By integrating multiplexed immunofluorescence with high-performance computing, we are learning how patterns of cellular phenotypes can predict clinical outcomes and act as a window to the mechanisms of cancer progression.


Many drugs exert their desired biological effects by interacting with multiple targets. In fact, analysis of drug-target databases shows that each approved drug interacts with an average of four targets. This raises the question whether polypharmacology is a necessary trait for efficacy of drugs. Phenotypic screens are naturally unbiased to any particular target, so they are able to discover compounds that may interact with a multitude of targets. To establish the basis for efficacy of compounds that we identify, we use cheminformatics and data-mining techniques to explore drug-target, activity, side-effect, and structural databases for target identification and side-effect prediction. By comparing HCS hit structures to compound structures and 3D shapes of annotated compounds and binding-sites of structurally resolved proteins, we identify potential targets. Testing and validation of these targets are done simply using their known modulators in the same phenotypic experimental setup. Once targets of hits are identified, we move on with computer-aided design and optimization of the compounds.


The ability to efficiently and accurately extract useful information from biological data allows us to maximize return on effort. Working in conjunction with other faculty of Biomolecular Informatics, we apply mathematical techniques such as probabilistic machine learning, artificial intelligence and Bayesian modeling to a variety of biological and pharamacological problems. Specific areas of focus include the analysis of combinatorial and statistical effects of drugs, predictive modeling of clinical outcome, data-based biomarker discovery and disease profiling, and translational bioinformatics.