Despite decades of research, there are no approved therapies to cure or even significantly slow AD progression post-diagnosis. Computational AD drug discovery studies, especially using both bioinformatics and cheminformatics approaches, have been very limited. However, there is an opportune time now to bring these approaches into AD field as large collections of experimental data are becoming available which relate many biological targets to AD pathogenesis and associate chemical structures to these targets. In parallel, observational data on AD relevance of various treatments from both epidemiological studies and social media including both general information exchange (Facebook or Twitter) as well as more specialized, health-related portals (e.g., WebMed), have emerged as well, along with rigorous text mining tools. This project will explore these orthogonal data streams concurrently and synergistically with the goal of identifying FDA-approved drugs that can be repurposed for AD prevention. We will employ statistical modeling of chemical genomics data for known AD targets, mine both social media and epidemiological data, and experimentally validate common drug repurposing hypotheses resulting from these analyses using both in vitro and animal models of AD. Specific Aim 1. Repurposing hypothesis generation based on chemogenomics data mining, curation, modeling, and target-based virtual screening: We will create an exhaustive list of potential AD targets by analyzing chemogenomics, transciptomics, and biomedical literature data sources. We will build cheminformatics models of these targets and employ these models for virtual screening of the FDA-approved drug library to identify compounds with AD-targeting potential. Specific Aim 2. Repurposing hypothesis generation based on observational data including epidemiological, social media, and medical literature sources: We will mine multiple social media sources (such as WebMD, Facebook, Twitter) for patient commentaries about unusual effects of drugs on cognition and memory improvement. Furthermore, using longitudinal clinical studies, we will perform epidemiological analyses to discern connections between infection (and its drug therapy) and increased AD risk. Specific Aim 3. Hypotheses fusion and experimentally validation of drug repurposing candidates in animal AD models: We will examine the existing drugs predicted to exhibit AD-target inhibition, and/or observed to show memory and cognition-enhancing properties, and/or neuroprotectivity (from studies in Aims 1 and 2) and prioritize a selection of hits for experimental validation in vitro and in animal models. The distinct feature of this project is integration of highly diverse sources of data on possible AD-relevant drug effects into a common data-analytical workflow and hypothesis fusion to arrive at a small number of promising candidates followed by the rigorous experimental confirmation. We expect that the unique combination of diverse expertise of coinvestigators and respective research tools in computational drug discovery, statistical epidemiology, and experimental AD studies will lead to the discovery of novel AD indications for existing drug therapies.

Today, large collections of experimental data are becoming available which relate many biological targets to AD pathogenesis and associate chemical structures to these targets. In parallel, observational data on AD relevance of various treatments from both epidemiological studies and social media have emerged as well, along with rigorous text mining tools. This project will explore these orthogonal data streams concurrently and synergistically with the goal of identifying FDA- approved drugs that can be repurposed for AD prevention.