PEOPLE: CURRENT LAB MEMBERS

Fig. 1: The statistical mechanical model for DNA unzipping. At one end of the double-stranded molecule, a “base” experiences pulling force F while the complementary “base” is fixed in space.

C. Brian Roland
Graduate Student
Department of Chemistry and Chemical Biology
Harvard University
12 Oxford St.
Cambridge, MA 02138
Tel: (617) 384-
Fax: (617) 384-9228
roland@fas.harvard.edu

DNA unzipping

Single-molecule experiments allow the direct measurement of the response of biopolymers to mechanical force. Theorists have shown that these experiments expose the mechanical consequences of heterogeneity in the sequence of monomer types [D. K. Lubensky and D. R. Nelson, Phys. Rev. E 65, 031917 (2002)]. In one class of experiment that addresses this issue, DNA is pulled apart in the unzipping configuration [C. Danilowicz et al., Phys. Rev. Lett. 93, 078101 (2004)]. In our work, we attempt to compute the temperature-force phase diagram of a
statistical mechanics model for DNA unzipping [Fig. 1]. The model includes both a random sequence of base-pairing energies and loops in the physical structure; these features distinguish our work from previous efforts. We use a new implementation of the replica method to address these features of the model.

Protein Structural Evolution

It is well accepted that two protein structural domains with high sequence-similarity probably descended from a common ancestor domain. But, when the pairwise sequence-similarity is low, the evolutionary relationship is unknown. In our work, we test the hypothesis that when the pairwise sequence-similarity is low, the structural-similarity between two structural domains is a true positive indicator of common ancestry.

Fig. 2: A portion of the graph for the network of structural relationships between domains present in Bacillus subtilis. [Figure reproduced with permission from Deeds et al].

We construct a model for the evolution of the network of pairwise structuralsimilarities between the structural domains present in an organism. The model contains only divergent mechanisms for the discovery of new domains. In particular, if two domains have high structural-similarity then they necessarily have a common ancestor domain. The model time-evolves a statistical ensemble of graphs, each graph representing a realization of the network of structural relationships. We compare the graphs generated by the model with “experimental” graphs, each graph made by comparing the PDB structures of domains present in a particular bacterial genome. We consider four bacterial organisms, each representing a major clade in a phylogeny based on structual domains as characters [see E. J. Deeds et al., Genome Res. 15, 393 (2005)]. We find that a typical model graph has a connectivity (degree distribution) similar to those of the “experimental” graphs.