Characterising novel insecticide transport to maximise efficiency and minimize environmental impact

全球最重要的优先事项之一是制定战略，支持为世界快速增长的人口提供高质量、有营养的食物。支持这一目标的一个重要环节是开发新型高效、安全、对环境影响小的杀虫剂。针对特定的活性蛋白位点，杀虫剂的开发已经取得了很大的成功，然而在杀虫剂分子的施用位点和活性位点之间存在许多物理和分子障碍。目前对活性分子和这些生物屏障之间发生的相互作用缺乏了解，事实上，通过这些屏障的运输效率不仅取决于活性杀虫剂本身，而且在很大程度上取决于应用配方及其物理状态。虽然在预测杀虫剂运输和功效的建模和机器学习方法方面取得了进展，但由于缺乏与膜运输相关的输入表征数据，这些方法也受到严重阻碍。在这个项目中，我们的目标是开发一个模块化的，基于实验室的，毫流体测量平台，它将允许快速表征杀虫剂跨系列模型膜和体外物理屏障的运输。这些数据将允许直接快速筛选和预测候选分子的效率，并优化配方，我们将输入机器学习方法，使我们能够基于分子结构预测膜运输。这里采用的方法将与巴斯夫内部使用的方法保持一致，因此将直接输入他们的监督和无监督预测模型，以显着提高杀虫剂发现和开发的效率和针对性。通过显著扩大发展参数空间，超越目标位点识别和相互作用，我们预计，在未来，这将导致杀虫剂配方的开发，具有更低的施用浓度，更高的特异性和更好的环境性能。

One of the greatest global priorities is finding strategies that will support the supply of high quality, nutritious food for the worlds fast-growing population. One vital strand that will support this is the development of novel efficient, safe insecticides with low environmental impact.The development of insecticides has been highly successful at targeting specific active protein sites, however there are a host of physical and molecular barriers between the application site and active site of insecticide molecules. There is currently a lack of understanding of the interactions that take place between the active molecules and these biological barriers and indeed, the efficiency of transport through these barriers is not simply based on the active insecticide its self, but depends strongly on the applied formulation and its physical state.While there has been progress in both modeling and machine learning approaches to predicting the transport and efficacy of insecticides, these approaches are also severely hampered by lack of input characterization data relating to membrane transport.In this project we aim to develop a modular, lab based, milli-fluidic measurement platform that will allow rapid characterisation the transport of insecticides across serial model membranes and in-vitro physical barriers. This data will allow direct rapid screening and prediction of efficiency of candidate molecules, and optimisation of formulation, and we will feed into machine learning approaches that will allow us to predict membrane transport based on molecular structure. The approaches employed here will be aligned with those used within BASF and so will feed directly into their supervised and unsupervised predictive models to significantly enhance the efficiency and targeting of insecticide discovery and development. By significantly expanding the developmental parameter space beyond target site identification and interactions, we anticipate that, in the future this will lead to development of insecticide formulations with lower applied concentrations, increased specificity and improved environmental performance.