Title: Adventures in Computational Biology: Modeling and Applications

Abstract:
With significant software and hardware advances, molecular dynamics (MD) simulations have become important for studying the motions of complex biological systems. For various DNA polymerases in the X family, classical as well as classical/quantum-mechanical simulations have uncovered conformational pathways to relate enzyme architecture to fidelity behavior. Going beyond MD, however, is necessary to capturing large-scale conformational changes and chemical pathways. Such methods include transition path sampling and coarse grained modeling approaches. For DNA polymerase beta, transition path sampling and hybrid classical/quantum approaches help relate free energy pathways to biological function. Studies of pol lambda and pol X elucidate the distinct pathways of these polymerases from each other and from pol beta. Applications to chromatin folding require a drastic reduction of the number of degrees of freedom by a coarse-grained approach. Using such a model of oligonucleosome chains in combination with tailored sampling protocols, we elucidate the energetics of oligonucleosome folding/unfolding and the role of each histone tail, linker histones, linker DNA length, and divalent ions in regulating chromatin structure. The resulting compact topologies reconcile features of the zigzag model with straight linker DNAs with the solenoid model with bent linker DNAs for optimal fiber organization.

Another area that requires innovative modeling tools involves RNA structure prediction and design. Our graph theory approach to represent RNA secondary structures, RAG (RNA-As-Graphs, http://monod.biomath.nyu.edu/rna), can be used to catalog all possible RNA 2D structure motifs as well as rank them by topological complexity. RAG has been used to classify/analyze topological characteristics of existing RNAs, analyze 3D RNA motifs, predict novel RNA motifs, and advance the design of novel RNAs by mimicking in silico the process of in vitro selection, which generally produces only simple topologies. Mimicking the experimental process can be done on the basis of a nucleotide transition matrix framework with supercomputing resources. Very large pools of nucleotides can be generated, screened, and filtered according to various 2D-structure similarity and flanking sequence analyses. The computational and theoretical yields for simple RNA motifs agree closely. For real aptamer targets, the in silico procedure overestimates the yields found experimentally, as expected, because experimental yields represent lower bounds and the screening does not yet involve 3D structural aspects.