Developing the Ecotoxicological - Predictive - Information - Connectivity Map (EPIC-map)

环境科学的一个基本挑战是建立暴露于化学应激源、其分子效应和由此产生的不良后果之间的关系。特别是在风险评估中，希望在其表现出来之前识别或预测不良后果。为此，计算（预测）生物学旨在开发技术和方法，以识别与不良结果直接相关的分子途径和生理终点。2003年，美国环境保护署（EPA）遵照白宫的命令，将计算预测生态毒理学置于其研发战略的中心，并从那时起就对这项技术进行了大量投资。在英国，NERC也通过建立环境组学合成（EOS）计划和将系统生物学作为核心研究重点之一，认识到集成多层次复杂数据集的计算方法的重要性。这些计划的一个重要组成部分是不良结果途径（AOP）框架，由美国环保局的G. Ankley等人开发。它的设计目的是将关键事件的可用知识链接到事件的逻辑链中，从分子启动事件到生物体分子和生理反应，并导致并包括不良结果。由此产生的路径可以被询问并用于支持风险评估和环境监测。随着aop数量的增加，可以对一种新化合物可能对特定生物体产生的影响做出更可靠的预测。AOPs的核心是重建导致不良结果的分子通路。这需要广泛和高度复杂的数据集，允许无偏的方法。因此，从这些数据中开发AOP所需的基础计算工具应该能够1)重建潜在的调控网络，2)将分子反应与不利结果联系起来，3)将化合物的结构特征与分子反应联系起来，以及4)为AOP和假设生成提供解释性输出。最后，从一组已知化合物中得出的结果构成了预测新化学物质对给定生物体的潜在分子效应的基础。该项目提案通过与美国环保署合作开发必要的计算框架来解决这一重要需求，该框架直接集成到其2015-17财年的研究计划中，以开发和验证计算（生态）毒理学技术，以整合并最终从传统分析中过渡。为了实现其目标，该项目将首先将各种公开可用的数据库与专门建立的200种化合物的标准化数据集整合到一个数据库中。基因和化合物之间的联系将构成网络方法的基础，而标准化数据集将用于识别目的和预测建模。为了查询数据库，该项目将利用许多计算生物学家目前使用的许多工具，此外还将纳入一套预测建模方法。这将允许任何工作台用户识别与其分子反应相关的化合物，用于AOP开发的基因-基因-化学-不良后果网络，以及通过化学结构对尚未测试的化合物的分子效应进行假设预测。这将为风险评估人员提供基础，如美国环保署、英国环境署或欧洲委员会联合研究中心，以及许多其他高层和学术组织，就化学品可能对生物体产生的潜在风险和影响提出假设。

A fundamental challenge within the environmental sciences is to establish the relationship between exposure to a chemical stressor, its molecular effect and the resulting adverse outcome. Particularly in risk assessment it is desired to identify or predict the adverse outcome before its manifestation. To this effect, computational (predictive) biology aims at developing techniques and methodologies to identify molecular pathways and physiological endpoints that can be directly linked to the adverse outcome. The Environmental Protection Agency (US EPA) following a White House Mandate in 2003 has placed computational predictive ecotoxicology at the centre of their R&D strategy and since then has been heavily investing into this technology. In the UK, the NERC has also recognised the importance of a computational approach to integrating multi-level complex datasets by establishing the Environmental Omics Synthesis (EOS) initiative and by including Systems Biology as one of the core research priorities. An essential part of the these programs is the adverse outcome pathway (AOP) framework, developed by G. Ankley et al at the US EPA.This was designed to link available knowledge on key events in a logical chain of events, from the molecular initiating event to organism molecular and physiology response and leading up to and including the adverse outcome. The resulting pathways could then be interrogated and used to support risk assessment and environmental monitoring. As the number of AOPs increase more robust predictions can be made on the effect a novel compound may have on a given organism. At the core of the AOPs is the reconstruction of molecular pathways leading to adverse outcomes. This requires extensive and highly complex datasets allowing for an unbiased approach. The underpinning computational tools required to develop an AOP from such data should therefore be able to 1) reconstruct the underlying regulatory network, 2) link the molecular response to the adverse outcome, 3) link structural features of compounds to molecular response, and 4) provide an interpretive output for AOP and hypothesis generation. Finally, the results developed from a set of known compounds forms the basis for deriving predictions on the potential molecular effect of novel chemicals on a given organism. This project proposal addresses this important need by developing the necessary computational framework in collaboration with the US EPA which directly integrates with their FY2015-17 research plan to develop and validate computational (eco)toxicology techniques to integrate and eventually transition away from traditional assays. To achieve its goals the project will start by integrating various publicly available databases with a purpose built standardized dataset of 200 compounds into a single database. Links between the genes and compounds will form the basis of the networking approach, while the standardized dataset will be used for identification purposes and predictive modelling. To interrogate the database the project will utilize a number of tools currently used by many computational biologists and in addition incorporate a set of predictive modelling approaches. This will allow any bench user to identify compounds linked to their molecular response, gene-gene-chemical-adverse outcome networks for AOP development and hypothetical prediction of molecular effect of yet untested compounds via chemical structure. This will provide a basis for risk assessors such as the US EPA, UK Environmental Agency or European Commission Joint Research Center and many other high level and academic organisations to develop hypotheses on the potential risk and effects a chemical may have on an organism.