My group continued to work on computational methods to study the dynamics of biological networks, impact of genetic variations and structural variation on gene expression, organismal phenotype and complex diseases. In addition we developed new methods for analysis of new high throughput experimental data. Specifically, we continued to work on methods to delineate genetic underpinning complex traits. Along this line continued to work on methods to detect epistatic interactions (1). We continued developing new methods to study the relation between genetic causes end the gene expression in cancer (2). We focused on approaches that help to delineate cancer heterogeneity. Our new approach based on topic models as been selected for presentation at RECOMB 2012 - a top computational biology conference and recently published in NAR (3). One important and novel aspect of this study is that unlike previous approaches which tried to enforce artificial boundaries between cancer subtypes, our new approach models individual cancers as mixtures of basic recurrent subtypes. Independently, we have pursued a different approach that leverage modularity of interaction network. The new method, termed Module Cover, allows identification of molecular sub-networks dys-regulated in cancer patients while accounting for disease heterogeneity (4). We also continued the collaboration with Brian Oliver's group on gene regulation and the impact of copy number variations in flies. Two papers resulting from this collaboration are currently in prepared for publication. In addition, we continued our research on DNA and RNA structures. In collaboration with David Levens and Rafael Casellas (equal contribution of all three groups) we have recently published a Cell paper reporting DNA melting as a regulatory step in resting B-Cell (5). We also worked with Leven's group on analysis of impact of transcription induced supercoiling (6). We continue the in-vivo studies of non-B-DNA structures. Focusing on more computational analysis, we reported an in-depth analysis of distribution of non-B-DNA susceptible sites in e.coli (7). In context of RNA structure, we have developed method to measure the impact of single nucleotide polymorphism on RNA secondary strucrure. Our method measures such effects as the relative entropy of Boltzmann ensembles of the native and mutated sequences. This approach has been proven to be very powerful (8). In the previous reporting period I reported development a computational approach to identify sequence/structure motifs of SELEX derived aptamers. This work has created tremendous interest and currently we collaborate whit a number of researchers on analysis of new experimental SELEX data and refining the method to work with HT-SELEX protocol. Finally, we continue to support other groups by providing computational expertise for their studies (9).
Showing the most recent 10 out of 72 publications
