home > research > sander research group

Sander Research Group

Publications

Computational biologists combine findings in biology with computer algorithms and databases to conduct biological research on powerful computers, using sophisticated software -- so-called "dry" laboratories -- in ways that complement and strengthen traditional laboratory and clinical research. The aim is to build computer models that simulate biological processes at different scales - from the molecular level up to the organism as a whole - and to use these models to make useful predictions.

Computational biology can help interpret detailed molecular profiles of cancerous and noncancerous cells, molecular response profiles of therapeutic agents, and a person's genetic profile to assist in the development of better diagnostics and prognostics, as well as improved therapies.

We are analyzing regulatory control in the TGF-b pathway by a combination of theory and experiment. Experiments involve systematic perturbation using RNAi in cell culture, with optical reporter constructs for downstream signaling and transcriptional activation, in a high-throughput screening facility. Collaboration with the Joan Massague's lab in the department of cancer biology and genetics and Hakim Djaballah's high throughput screening facility.

To discover gene sets and processes characteristic for cancer subtypes and indicative of prognosis and response to therapy, we map DNA microarray data to pathway and interaction network for visual and algorithmic processing. An application is the development of characteristic transcript signatures for custom arrays suitable for clinical applications. Collaboration with oncologists and pathologists.

To predict sites of functional interactions on protein molecules and elucidate the specificity structure in interaction networks, we apply an entropy-based algorithmic analysis of protein family sequences. Specific applications are to the interpretation of large-scale binding experiments on PDZ domains and to the interpretation of genetic variation in a set of candidate genes sequenced in melanoma. Collaboration with epidemiologists at MSKCC and proteomics groups at Genentech.

To help discover the currently largely unknown targets of microRNA regulation of transcription, we apply a dynamic programming algorithm on a genome-wide scale to microRNAs from flies to humans. Close collaboration with RNA group at Rockefeller U.

Computational screening of small-RNA profiles (from cloned libraries) led to the discovery of microRNA genes in herpes viruses, including the Kaposi Sarcoma virus. This has opened a new chapter in virology and AIDS research. Target discovery is in progress. Close collaboration with Tom Tuschl's group at Rockefeller U.

We are currently engaged in a systematic bottom-up approach to the analysis and simulation of networks, covering these levels:

Bottom - Component Properties
Data	Capture pathways of interest in a database, currently focusing on regulatory pathways in cancer, represented as genes, gene products and typed interactions, including reactions
Knowledge	Develop ontology for biological networks (bioPAX) as the basis for a standard format for the exchange of pathway information; Merge in-house and external pathway data into an information system
Tools	Build visualization and analysis software workbench in an open-source collaborative effort (Cytoscape, 4 groups) and develop algorithmic plug-ins, e.g., activity center algorithm (Pradines et al., Sander, 2004)
Analysis	Develop algorithms for motif detection, motif clustering and motif similarity searches; analyze evolutionary dynamics
Simulation & Prediction	Simulate selected pathways (sub-networks), predict the effects of perturbations, derive regulatory network characteristics, such as adaptation, robustness, and type of feedback control, design & interpret experiments
Application	Applications to different areas of biology; currently developing systems approach to multi-agent intervention for therapy
Top - System Properties