Research Biclustering / Transparent Probabilistic Data Integration: Grouping genes into functionally related and putatively co-regulated clusters is an essential first step for the inference of regulatory networks (one could think of many reasons for doing this but network inference is a good problem to start with). It is widely known that regulatory relationships among genes can vary under diverse environmental settings, and that co-expressed genes are trivially under control of the same regulator(s). This leads to patterns of co-expression that are valid under some, not all, observed conditions. With such considerations in mind, we have developed cMonkey, an unsupervised learning procedure for detecting putatively co-regulated gene clusters by integrating diverse systems biology data including: (1) mRNA and/or protein expression levels, (2) cis-regulatory sequences, and (3) functional association and physical interaction networks. Network inference: We have also developed a methodology for deriving transcriptional regulatory interactions on a genomewide scale, and apply the method to predict a large portion of the gene regulatory network of the archaea, Halobacterium NRC-1. A small portion of the halobacterium learned network is shown below. The learned network is predictive, and was used to successfully predict the global expression of Halobacterium under novel perturbations (not part of the original training set) with predictive power similar to that seen over the training set. Methodological advancements over earlier work include an explicit treatment of time such that the network model can be fit using both steady-state measurements and heterogeneous time series simultaneously. The method contains a novel means for learning binary logic interactions between regulators that requires no discretization of data. Rosetta de novo structure prediction: methods development for extracting functional information fro de novo structure predictions. Recent progress in de novo structure prediction methods has resulted in methods with increased accuracy that are applicable to greater numbers of proteins. When combined intelligently with other structure prediction methods, de novo structure prediction can contribute to systems biology in several ways. While still highly experimental such applications include 1) structural annotation on a genome wide scale and 2) synergy with experimental approaches to structural genomics such as the derivation of distance constraints from mass spectroscopy. I will describe the underlying methodologies common to current de novo prediction methods, focusing on core concepts rather than specific implementations, groups or methods. Possible applications of de novo structure prediction will also be reviewed. For more information, view our latest results on 80 complete genomes including many model organisms being actively studied at NYU. This work is being carried out in collaboration with David Baker. Cytoscape: Cytoscape is a computer program designed to visualize systems biology data. This began initially as a program to map expression data onto networks and has evolved parallel to the field to encompass several other functionalities. We are involved in cytoscape development with a focus on providing immersive tools that can be used to view the data created by the HPF project (in collaboration with Iliana Avila-Campillo at the ISB). BioNetBuilder is a tool developed by Iliana to automatically build networks in cytoscape for any organism comprised of edges derived from protein-protein, protein-DNA, evolutionary comparison and metabolic/signaling pathways. For more information on Cytoscape see cytoscape.org. Areas of Research/Interest Systems Biology and Protein Modeling Fellowships/Honors Howard Hughes Medical Institute pre-doctoral Fellowship in the Biological Sciences, 1998-2001; Magna Cum Laude, Biochemistry, FSU, 1996; American Cancer Society – James Jay Fisher Fellowship, 1996; Florida Academic Scholars Award, 1993; International Baccalaureate Degree, 1993.
Publications
|
|
|
