| Our aim is to increase the rate with which accurate and relevant biological and chemical results are elicited from the ever expanding corpus of experimental data and analytical techniques. In particular, we have developed a series of accurate and efficient algorithms capable of assisting the biologist and chemist in deciphering experimental data. We have developed efficient algorithms for deterministically identifying differences among mass spectrometry proteomics data, modeling protein flexibility for drug design and protein redesign using molecular ensembles, augmenting molecular replacement techniques for generating initial phases in X-ray crystallography, and utilizing homologous structural information for automated backbone resonance assignment in Nuclear Magnetic Resonance (NMR) spectroscopy.; The development of algorithms in computational biology is important to both the biological and computational communities. From the biological and chemical perspective these new algorithms may extract previously unavailable information from existing or novel data sources. This information may assist the discovery process by allowing one to pose and answer novel biological and chemical questions. Furthermore, efficient computational algorithms may reduce both the amount and the cost of biological and chemical wetlab testing required to obtain these answers. In other words, each algorithm in computational biology should be considered in the same light as any new wetlab experimental technique; these algorithms will provide unique information, novel analysis, and key insights to the biological problem at hand. From the computational perspective, many of the chemical and biological problems attacked by computational biology present as algorithmic challenges in computational geometry, information theory, linear algebra, graph theory, and machine learning. To maximize the chance of providing useful analysis, algorithms in computational biology should be provably correct under a biologically reasonable set of criteria. If possible, these algorithms should be efficient and run in polynomial time; however, because polynomial-time algorithms may not be attainable for many biologically and chemically interesting problems, the computer scientist is often challenged to develop approximation algorithms capable of producing results in a manageable amount of runtime. To be useful, algorithms in computational biology must generally be robust to noise, experimental error, and missing data. Developing such rigorous algorithms in any context remains a significant challenge. The field of computational biology thus offers both significant challenge and potential benefit to both the biological and computational communities. |
