Integrative computational biology

目的：整合计算生物学是深入理解复杂疾病生物学、实现个体化医学的必要平台。然而，孤立地推进计算工具不足以影响计算生物学。由于问题的规模或复杂性，或者由于使用不切实际的假设，许多理论上优秀的方法对于高通量生物学领域是不够的。形成假设和开发生物系统的计算模型，而不可能对其进行测试，限制了获得现实模型的潜力。我们建议提高模式发现算法的可扩展性鲁棒性，灵敏度和特异性，并将它们集成到系统生物学分析和癌症研究中高通量数据的可视化的方法中。 方法：为了解决癌症信息学中的挑战，我们建议：1）将多个数据资源整合到生物数据集市中; 2）设计用于系统的、基于网络的异质生物数据分析的工具; 3）将这些工具应用于上皮癌建模，并通过合作启动验证研究。我们将使用统计和机器学习算法来增强工具，以实现概率推理，确定性和声音模式识别，稳定性，可扩展性和鲁棒性。算法开发有三个主要领域：1）癌症谱的聚类和数据挖掘算法; 2）蛋白质相互作用网络的图论算法;以及3）网络背景下结果的集成和可视化，以帮助假设生成，结果解释和模型创建。 意义：我们的目标是开发集成和可扩展的算法，并应用它们来实现智能分子医学。这些分析的结果将被整合到计算模型中，这将提高其准确性和适用性，进而影响各个领域。这项建议的一个重要作用是培训生物信息学专业人员，而这方面的人才仍然严重不足。我们将发布免费学术使用的工具和资源。

Objective: Integrative computational biology is a necessary platform to fathom biology of complex diseases and impact individualized medicine. However, advancing computational tools in isolation is insufficient to impact computational biology. Many theoretically excellent approaches are inadequate for the high-throughput biological domains, due to the scale or complexity of the problem, or due to using unrealistic assumptions. Forming hypotheses and developing computational models of biological systems without the possibility to test them limits the potential to derive realistic models. We propose to improve scalability robustness, sensitivity and specificity of pattern discovery algorithms, and integrate them to support a methodical approach to the systems biology analysis and visualization of high-throughput data in cancer research.   Approach: To address challenges in cancer informatics, we propose to: 1) integrate multiple data resources into biological data marts; 2) design tools for systematic, network-based analysis of heterogeneous biological data; 3) apply the tools to modeling epithelial cancers and initiate validation studies via collaborations. We will augment the tools with statistical and machine learning algorithms to enable probabilistic reasoning, deterministic and sound pattern identification, stability, scalability and robustness. There are three main areas of algorithm development: 1) clustering and data mining algorithms for cancer profiles; 2) graph theory algorithms for protein interaction networks; and 3) integration and visualization of results in the context of networks, to aid in hypotheses generation, interpretation of results, and model creation.      Significance: We aim to develop integrated and scalable algorithms, and apply them to realize intelligent molecular medicine. Results from these analyses will be integrated into the computational models, which will improve their accuracy and thus applicability, and in turn will impact in the respective fields. An important function of this proposal is the training of bioinformatics professionals for which there is still a severe deficit. We will release tools and resources for free academic use.