Center for High-Performance Computing and Communications

Remo Rohs


Remo Rohs is an assistant professor of biological sciences, chemistry, and physics in the USC Dana and David Dornsife College of Arts, Letters, and Sciences. Rohs and his team strive to integrate two key aspects of DNA research, those of sequence and structure.

When scientists analyze genomes from the perspective of  sequencing, they are looking for the one-dimensional placement of the four possible nucleotides—labeled A, C, G, and T—within a given sample of DNA. Structural analysis, meanwhile, generates three-dimensional representations of these complicated chemical units. There is a nontrivial relationship between DNA sequence and shape. For instance, shape helps determine how accessible a given base pair might be for protein contacts. A clearer grasp of how transcription factors bind DNA could have significant applications for drug design and cancer therapy.

Rohs is at work on creating computational methods for the high-throughput prediction of DNA shape. Supercomputing plays an integral role in analyzing the massive data sets required to predict local DNA shape of whole genomes. The underlying data for the genome-wide shape prediction is generated in computationally extensive all-atom Monte Carlo simulations. Rohs and his team have performed thousands of these simulations on the HPCC cluster. The ability to make DNA shape predictions in a high-throughput manner will have significant impacts on genome analysis.

Rohs recently published a paper on this new method, developed by him and his group, in the journal Cell. His other research interests include nucleosome positioning and the effects of chemical modifications, such as DNA methylation and base pairing variants, on DNA shape, as well as on protein-DNA recognition. Rohs’s research is currently funded by a grant from the American Cancer Society, the USC-Technion Visiting Fellows Program, and an Andrew Viterbi fellowship.

ABOVE: Eight Drosophila Hox proteins bind to very similar target sites but execute distinct in-vivo functions. The figure illustrates that the cofactor Extradenticle (Exd) (yellow) unlocks the specificity of Hox proteins (cyan) for recognizing DNA target sites (metallic). Specificity fingerprints of Hox proteins reveal that DNA shape is a determining factor in achieving specificity.