Advanced probabilistic model building genetic algorithms and application to bioinformatics

实际应用程序的复杂性和/或待分析数据的大小使得很难在合理的计算时间内找到问题的精确解决方案。在生物信息学领域尤其如此。生物模型非常复杂，由基因组测序、蛋白质组学、基因调控、蛋白质/蛋白质相互作用以及现在的宏观生物学（细胞模型、生物体或生态系统模拟等）项目产生的数据通常呈指数级增长。因此，提供有效的启发式方法来分析此类问题是非常重要的。Gras提出了一种基于概率引导启发式优化策略的研究方案。这是最近出现的启发式策略的一个新领域。它是基于强大的技术从统计分析和组合优化。这是一种创新和有前途的方法，因为生物信息学问题的主要部分可以表示为优化问题。基于概率模型构建的现代元启发式方法已经证明了其解决这类优化问题的潜在效率。此外，由于构建了概率模型，这些方法可以揭示关于数据内在结构的非常丰富的知识。Gras博士建议应用这些方法并设想新的方法，结合使用专家（生物学家）知识来解决生物信息学问题。由于该计划旨在开发工具来分析与阿尔茨海默病、心血管疾病、牛海绵状脑病和癌症相关的生物数据，因此可以更好地理解这些现象，并发现新的诊断技术，甚至新的治疗方法。该方案还包括分析，然后建立一个模型和生态系统的模拟。这对于理解生物种群的进化和污染对生态系统的影响至关重要。

The complexity of real applications and/or the size of the data to be analyzed makes it difficult for exact solutions to problems to be found in reasonable computational time. This is especially true in bioinformatics. Biological models are extremely complex and the data generated by genome sequencing, proteomics, gene regulation, protein/protein interaction and now macro-biology (cell models, organism or ecosystem simulation, etc.) projects often grow exponentially. Providing efficient heuristics to analyze such problems is therefore very relevant.Dr. Gras propose a research program based on optimization strategies using probabilistically guided heuristics. It is a new domain of heuristic strategies that has emerged recently. It is based on powerful techniques from statistical analysis and combinatorial optimization. This is an innovative and promising approach as a major part of the bioinformatics problems can be represented as optimization problems. Modern metaheuristics, based on the building of a probabilistic model, have proven their potential efficiency to solve such optimization problems. Moreover, due to the construction of the probabilistic model, these methods can reveal very informative knowledge about the intrinsic structure of the data. Dr. Gras proposes to apply such approaches and conceive new ones, associated with the use of expert (biologist) knowledge to solve bioinformatics problems.As this program aims to develop tools to analyze biological data that are correlated to Alzheimer disease, cardiovascular disease, bovine spongiform encephalopathy and cancer, it can lead to a better understanding of these phenomena and to the discovery of new diagnostic techniques and even new therapies. This program also involves the analysis and then the establishment of a model and the simulation of ecological systems. That would be essential for the understanding of evolution of biological populations and of pollution effects on ecosystems.