Super-resolution microscopy network analysis: Drug target validation for cystic fibrosis

基于基因组的精准医学的目标是选择性地将药物靶向有反应的患者。这在囊性纤维化（CF）中尤其重要，其中约75%的携带F508 del-CFTR突变的合格患者在使用临床可用药物Orkambi后未显示出肺功能的明显改善。Orkambi治疗费用> 250，000美元/年，目前在加拿大没有正式由医疗机构资助的医疗保健覆盖。2期临床试验中的新药可能比其他药物对个体患者更有效。为了确定哪些患者最有可能从CF调节剂药物中获益，重要的是在蛋白质水平上研究CFTR，因为患者之间的差异不仅取决于基因组差异（例如特定的CFTR突变）。我们将应用基于机器学习的单分子定位超分辨率显微镜（SMLM）网络分析来开发一种基于成像的检测方法，以识别可能从特定治疗中受益的患者。优化个体患者的治疗将提高药物疗效，并大大降低加拿大CF患者的治疗费用。SMLM产生的分辨率超过衍射极限十倍。然而，定量的方法来分析所产生的三维点定位的大数据集是有限的。我们开发了一种新颖而强大的方法来建模SMLM粒子定位数据集，通过点之间的虚拟连接将3D定位转换为网络。这种方法使我们能够确定分子结构，即蛋白质复合物中的蛋白质如何相互作用。我们现在建议改进该技术并将其应用于CFTR，从而验证我们的分析管道在CF和其他具有已知药物靶点的疾病中的个性化药物应用。我们将与加拿大显微镜公司Quorum Technologies合作，将该软件商业化，供全球超分辨率显微镜用户和制药公司使用。

The goal of genome-based precision medicine is to selectively target drugs to responsive patients. This is particularly important in cystic fibrosis (CF), where ~75% of eligible patients, carrying the F508del-CFTR mutation, do not show demonstrable improvement in lung function with the clinically-available drug Orkambi. Orkambi treatment costs >$250,000/yr and is not formally covered by publically funded health-care in Canada at this time. New drugs in phase 2 clinical trials may be more effective than others for individual patients. To identify which patients are most likely to benefit from CF modulator drugs, it is important to study CFTR at the protein level as patient-patient variation is not only dependent on genomicdifferences (e.g. specific CFTR mutations). We will apply machine learning based network analysis of single molecule localization super-resolution microscopy (SMLM) to develop an imaging-based assay to identify patients likely to benefit from a particular therapeutic. Optimization of therapies for individual patients will increase drug efficacy and substantiallyreduce cost of treatment for Canadian CF patients. SMLM generates resolution ten-fold beyond the diffraction limit. However, quantitative approaches to analyze the large data sets of 3D point localizations generated are limited. Wedeveloped a novel and powerful approach to model SMLM particle localization data sets, transforming 3D localizations into a network by virtual connections between the points. This approach allows us to determine molecular architecture, i.e. how proteins within a protein complex interact. We now propose to refine and apply the technology to CFTR and thereby validate our analysis pipeline for personalized medicine applications in CF and other diseases with a known drug target. We will partner with Quorum Technologies, a Canadian microscopy company, to commercialize the software for use by super-resolution microscopy users and pharmaceutical companies worldwide.