High Performance Computing In Molecular Biology

Sequences of DNA and RNA make up the genomes of organisms. As sequences accumulate we see an increasing role for computation in utilizing this data for a deeper understanding of all living things. The genomic era started with the revelation that the arrangement of four chemical subunits of nucleic acids in long sequences gave the information for the sequential arrangement of the 20 chemical subunits of proteins. As complete genomes of many viruses, microbes and parts of higher organisms appear it is obvious that the even the human genome is likely to be determined in less than a decade. Computational challenges arise fkom our immediate need to understand not only the linear data of sequences but the 3dimensional constellations of atoms in the molecules that they represent. All biological bctions are expressed through a rich variety of molecular interactions. Enzymes perform catalytic roles by interacting with chemical bonds. Regulatory and structural functions are expressed through molecular associations generally involving non-covalent bonds. Consequently, knowledge of structures is vital to discerning functions. Increasingly powerful computers ' can simulate these properties, sometimes giving insight into details that are difficult to obtain experimentally. This paper addresses several immediate areas of opportunity for computation involving genomic data. Sequence data acquisition, storage, retrieval and analysis are areas of intense activity. These activities are oRen called bioinformatics. The employment sections of scientific journals reflect the need of industry and academia for computational people to be central players, along with experimentalists, in many projects. Another area involves the analysis of biophysical data fiom crystallography or spectroscopy to produce high resolution structural details. Coming fiom biophysical foundations scientists in this area embraced high performance computing before many others in the biological field. An area of even more intensive computational challenge is given by the availabjlity of high resolution structural data. Computation can add value through refinement and analysis. Even more tantalizing is the prospect. for the use of computers to explore and to discover the principles of macromolecular structure, which may then be used to derive rules for