About Alzheimer's Disease Workbench

Introduction

The goal of Alzheimer's Disease Workbench (ADW) is to develop a multi-tier architecture with all the data and resources accessible through a cloud environment. The server will be housed at the San Diego Supercomputer Center. The data from Target Molecule Pages, the multi-omics measurements on model systems, assays performed in the Center and the target lead molecules will all be entered into a PostgreSQL relational database system. All tools for data analysis including modules using “R” statistics packages, omics functional and pathway annotations, quantitative modeling programs and target perturbation analysis tools, will be made available with easy-to-use interfaces on the ADW.

At present, we have R shiny app for RNA-seq analysis and we also provide a detailed protocol to do end-to-end analysis of RNA-seq datasets

A schematic of the ADW is presented here.

Alzheimer's Disease Workbench Web Portal

The ADW will serve as an entry point to all the resources that are publicly available and those developed in our Laboratory at UC San Diego. All AD databases in ADW and links to all publicly available data resources can be found here . Other resources that are available from us- tools, APIs and User Interfaces that will facilitate query, download, analysis and visualization of data from the Center and other data sources; an experimental tracking system using iPython and Jupyter Notebooks that will be FAIR and Cloud compliant; and educational and outreach material will be made available soon.

ADW Databases

All publicly available databases working in direction of Alzheimer's disease are available here . In addition, one of the goals of ADW is the creation of a query engine on our portal that will facilitate cross querying and where feasible provide integrative data analysis. In addition to the above, AD research continues to develop extensive molecular profiling of human subjects and model systems from transcriptomics, proteomics, metabolomics and other phenomics measurements. While some of these data are captured in databases mentioned above, it is difficult to interoperate and integrate data for analysis. The tools and APIs provided in ADW will facilitate this vertical integration of data. All metadata, data, protocols and workflows will be made available soon through iPython and Jupyter Notebooks to all users.

ADW Target Molecule Pages

The ADW Target Molecule Pages will contain comprehensive data/annotations about all the targets that have been identified and will continue to be dynamically updated. It will include automated annotation data on each target, including known polymorphisms (from public databases and human genome sequence data), all annotated functional data including GO annotations, pathway and disease annotations, and cellular and expression data. The Molecule Pages will also present data pertaining to disease relevance; and functions, pathways and endotypes. The Target Molecule Pages will also contain query interfaces that will facilitate posing simple text or complex Boolean queries of the database. The entire target molecule data set including all annotations will be downloadable by the community. This page is under development and will be made available soon.

ADW Resource Center

Our laboratory continues to develop significant resources including model iPSC and iN cell systems, reagents including antibodies, shRNA and CRISPR constructs, probes, etc. These resources will be presented on ADW along with annotations and clear instructions on mechanisms of access to the public.

ADW Tools/APIs

• Statistical Analysis Tools. We have developed tools for performing analysis of diverse data. Resources include principal component analysis and discriminant analysis procedures in both supervised and unsupervised mode, various hierarchical cluster analysis and correlation methods as well as identification of significantly modulated molecules by volcano plot analysis and testing for sample integrity with relative log abundance plots. Customized sets of online tools will be uploaded on ADW. Data at the individual molecule level will be analyzed by boxplot, bar graph, jitter plot, ANOVA, t-test and Z-score methods.
• Omics Tools. We present multiple analytical tools for “omics” data. All tools mentioned there will continue to be made available on or through links from ADW.
• Pathway Tools. A curated pathway database that contains extensive knowledge of existing human pathways is a central component to aiding and accelerating the interpretation of AD target data. We will make available publicly available pathway tools on ADW.
• APIs. Over the past decades, the broader research community has developed several useful tools for omics. A challenge for the AD research community will be to integrate these tools with the proposed infrastructure (data, interfaces and other tools). In order to accomplish this task, we will actively develop Application Programming Interfaces (APIs) using web service protocols, such as REST. These will enable a) access of newer tools from ADW to the community, and b) an individual researcher or a research site to integrate their tools dynamically with the data and tools available on the ADW website.

ADW Web-based Query

The web-interface layer will provide simple means by which the user will interact with the resource. The web interface will allow users to enter queries in plain English, and to construct complex questions using various logical operators and biological/clinical terms, e.g., SELECT all pathways AND all model systems affected AND current drugs available when endotypes include hyperphosphorylated tau protein and dementia. A back-end analytics engine will be implemented in R/python. For NLP, we will use RopenNLP interface. A combination of PHP/PostgreSQL will be used for detailed query on the databases. Graphical visualization will be facilitated through graphics elements of R and Javascript Libraries such as JS Charts and JQuery. he ADW resource will be powered by a high performance-computing infrastructure maintained by San Diego Supercomputer Center with more than 170 computing nodes with over 3000 processing units.