High performance computing for computational Biology

我们计算生物学小组的研究重点是开发和应用高效的计算技术，以解决生物医学科学(有机化学、生物化学和生物学)中的数据分析和建模问题。我们的实验室旨在生产在一组生物合理的标准下被证明是正确的算法，并且对实验噪声和丢失的数据具有健壮性。我们研究有机分子，从DNA(生命的分子蓝图)，到小配体(具有数十个原子的药学有趣的分子)，再到大的蛋白质(具有数千个原子的巨大分子)。我们感兴趣的是表征这些分子的结构和功能，然后利用这些信息帮助我们的生物合作者开发新的药物和疾病模型。在计算机科学中，问题通常通过采用一长串简化假设来净化，这些假设的集合允许开发极其高效的算法。然而，当我们开发用于科学计算(如化学和生物)的算法时，我们通常无法在不损害结果的科学有效性的情况下做出这些假设。因此，我们开发高效算法的最佳尝试往往会产生在标准桌面硬件上运行仍需要数周时间的程序；因此，大多数计算生物学实验室，包括我们的实验室，都利用计算集群形式的高速计算资源来更快地获得结果。该计划将通过提供具有大容量磁盘存储、磁带备份和安全VPN连接的高性能计算群集，帮助为我们实验室的计算基础设施提供资金。

Research in our Computational Biology group centers around the development and application of efficient computational techniques to solve data analysis and modeling problems in the biomedical sciences (organic chemistry, biochemistry, and biology). Our lab aims to produce algorithms that are provably correct under a biologically reasonable set of criteria and that are robust to experimental noise and missing data. We study organic molecules from DNA (the molecular blueprint of life), to small ligands (pharmaceutically interesting molecules with tens of atoms), to large proteins (enormous molecules with thousands of atoms). We are interested in characterizing the structure and function of these molecules and then utilizing this information to assist our biological collaborators in the development of new pharmaceuticals and models of disease. In Computer Science, problems are generally sanitized by adopting a long list of simplifying assumptions, the collection of which allow development of extremely efficient algorithms. However, when we develop algorithms for scientific computing (such as chemistry and biology) we are generally unable to make these assumptions without damaging the scientific validity of our results. Consequently, our best attempts at developing efficient algorithms often produce programs that still require weeks to run on standard desktop hardware; therefore, most computational biology labs, including ours, utilize high-speed computing resources in the form of compute clusters to obtain results more quickly. This proposal will help fund the computational infrastructure of our lab by providing a high-performance compute cluster with large disk storage, tape backup, and secure VPN connection.