Machine Vision Analysis of C. Elegans Phenotypic Patterns

描述（由申请人提供）：复杂的基因网络是人类疾病和健康的基础。基因网络的构建现在是酵母和培养的哺乳动物细胞等简单细胞的标准技术。多细胞生物的网络推理尤其有希望，但一个挑战是将网络解析为功能通路，而不仅仅是连接的图，第二个挑战是分析复杂表型（如神经元功能和行为）的网络。我们的目标是用秀丽隐杆线虫作为模型来学习如何完成这项任务，同时生成一个将为人类遗传学提供信息的网络。特别是，我们将继续利用我们的半自动运动分析系统（WormTracker）来获得大量基因的表型谱。基因将使用可用的功能丧失突变被询问。检查的基因将包括所有相关的神经元基因，以及编码染色质修饰蛋白和转录因子的基因。转录调控因子或染色质修饰蛋白与神经元效应基因的计算聚类将推断基因之间的调控关系。除了进食时的运动，我们也会记录进食时的运动，包括爬行和游泳时的运动。我们将聚类表型来推断遗传模块，并使用其他可用的基因组规模数据（如基因表达数据）扩展这些模块。为了获得药物-基因网络，我们将分析一组具有代表性的药物，并将它们与基因表型谱进行比较。我们将通过测试特定的药物-基因相互作用来测试药物-基因网络的预测。为了完善遗传网络，我们将开发额外的表型分析方法，并将其应用于基因、药物和基因-药物相互作用，以将网络划分为表型空间区域。这些试验将包括使用微流体装置对咽泵的速率和变化进行定量、自动分析，建立对男性尾部姿态和针尖延伸的药理学影响的试验，以对更复杂的男性神经系统进行样本遗传影响，以及化学引诱剂和驱避剂小组来监测感觉反应。我们将通过整合从WormBase导入的大量定量行为表型数据集和现有信息来利用我们的结果，这些信息允许遗传网络推断（表达数据、体外结合、基因本体注释、染色质免疫沉淀数据等）。硬件建设的软件和协议将在实验室网站上免费提供。

DESCRIPTION (provided by applicant): Complex genetic networks underlie human disease and health. The construction of genetic networks is now a standard technique in simple cells such as yeast and cultured mammalian cells. Network inference for multicellular organisms is promising especially but one challenge is to parse the network into functional pathways as opposed to just connected graphs, and a second challenge is to analyze networks for complex phenotypes such as neuronal function and behavior. Our goal is to use C. elegans as a model to learn how to accomplish this task, meanwhile generating a network that will inform human genetics. In particular, we will continue to exploit our semi-automated locomotion analysis system (WormTracker) to obtain a phenotypic profile for a large set of genes. Genes will be interrogated using available loss-of- function mutations. The genes examined will include all relevant neuronal genes, as well as genes that encode chromatin modifying proteins and transcription factors. Computational clustering of transcriptional regulators or chromatin modifying proteins with neuronal effector genes will infer regulatory relationships among genes. In addition to locomotion on food, we will also score locomotion off food, and both during crawling and swimming. We will cluster the phenotypes to infer genetic modules, and expand these modules using other available genome- scale data such as gene expression data. To obtain a drug-gene network, we will profile a representative set of drugs and compare them to gene phenotypic profiles. We will test predictions of the drug-gene network by testing particular drug-gene interactions. To refine the genetic network, we will develop additional phenotypic profiling methods, and apply to genes, drugs and gene-drug interaction to split the network into regions of phenotype space. These assays will include quantitative, automated analysis of the rate and variation in pharyngeal pumping using microfluidic devices, established assays for pharmacological effects on male tail posture and spicule protraction to sample genetic effects on the more complex male nervous system, and panels of chemoattractants and repellants to monitor sensory responses. We will leverage our results by integrating what will an extensive data set on quantitative behavioral phenotypes with existing information that allow genetic network inference (expression data, in vitro binding, Gene Ontology annotations, Chromatin immunoprecipitation data, etc.) imported from WormBase. Software and protocols for hardware construction will be freely available from laboratory websites. PUBLIC HEALTH RELEVANCE: Complex genetic networks underlie human disease and health but are a challenge to elucidate. We will use the model organism C. elegans to elucidate genetic networks underlying behavior by efficiently obtaining quantitative behavioral data on mutant strains that are defective in single genes using automated, machine vision systems. The quantitative data will be used to computationally infer genetic networks including genes that function in the nervous system and those that regulate other genes. The data and inferences will be publically available through the Neuroscience Information Framework and WormBase; the software for machine vision will be freely available for download.