About DataLENS

Alzheimer DataLENS is an open data analysis platform which aims to advance Alzheimer’s disease (AD; see About Alzheimer’s Disease below) research by enabling the analysis, visualization, and sharing of -omics data. DataLENS houses bioinformatics pipelines for the analysis of -omics data on Alzheimer’s disease and related dementias (ADRD) as well as streamlined web interfaces which allow neuroscientists to browse and query the results of these analyses. Note that -omics refers to the branch of science concerned with quantifying the levels of biological molecules (e.g., RNAs, proteins, etc.), and encompasses genomics, transcriptomics, proteomics, metabolomics, metagenomics, etc.

There are currently over 50 genetic, proteomics, and transcriptomics studies included in the DataLENS database, most of which are already available to analyze and explore:

Gene Expression Data

1. Analysis of 60 human microarray expression profiling datasets across various neurodegenerative diseases (26 Alzheimer’s, 21 Lewy body dementia, 13 amyotrophic lateral sclerosis and frontotemporal dementia).
2. Analysis of 30+ public human datasets spanning 19 brain regions and 5 cohorts.
3. Analysis of data from several Alzheimer’s disease animal models.
4. Three single-cell RNA-sequencing datasets.

Proteomics Data

1. Analysis of two proteomics studies, with additional studies currently in progress.

Genome-Wide Association Studies (GWAS)

1. Results from the International Genomics of Alzheimer’s Project (IGAP) GWAS meta-analysis.
2. Results from the Accelerating Medicines Partnership – Alzheimer’s Disease (AMP-AD) GWAS study.

Pathway Analysis

1. Protein-protein interaction data and integration of expression, epigenetic, and genetic data.

Installation

DataLENS is an R Shiny web application with an HTML/CSS/JavaScript front-end, an R back-end, and a MongoDB database.

Development

To install the development version of DataLENS, please complete the following steps:

1. Install the open-source R programming language and statistical computing environment, available here. DataLENS has been tested with R version 4.1.1 (2021-08-10, “Kick Things”) on Windows 10 (x86_64-w64-mingw32/x64).
2. Install the RStudio integrated development environment (IDE), available here.

Next, please install the required dependencies in R:

Package	Version
shiny	>= 1.7.1
shinydashboard	>= 0.7.1
mongolite	>= 2.4.1
AnnotationDbi	>= 1.54.0
org.Hs.eg.db	>= 3.13.0
httr	>= 1.4.2
data.table	>= 1.14.0
purrr	>= 0.3.4
magrittr	>= 2.0.1
ggplot2	>= 3.3.5
plotly	>= 4.9.3
DT	>= 0.20
ggseg	>= 1.6.4
igraph	>= 1.2.6
ggraph	>= 2.0.5
ggiraph	>= 0.7.10

Most packages can be installed from The Comprehensive R Archive Network (CRAN) as follows:

install.packages("<package-name>")

Multiple packages can also be installed simultaneously; see ?install.packages() for more information.

The AnnotationDbi and org.Hs.eg.db packages are available from the Bioconductor project. To get the latest release of Bioconductor followed by these packages, please run the following:

# install Bioconductor
if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install(version = "3.14")

# install packages
BiocManager::install("AnnotationDbi")
BiocManager::install("org.Hs.eg.db")

MongoDB Installation and Database Import

DataLENS employs a MongoDB back-end database. Note that the following steps assume a Windows operating system; however, analogous instructions for macOS or Linux can be found here.

Installation

To install MongoDB locally, please download MongoDB 5.0 Community Edition, available here. Follow the steps provided by the MongoDB Installer wizard. For production on a remote server, it is recommended to install MongoDB as a Windows service (i.e., an application that runs in the system’s background; see here). For development, it may be preferred to just install the binaries; however, note that a MongoDB instance will have to be manually launched each time DataLENS is run.

It may also be advantageous to install MongoDB Compass (i.e. the MongoDB graphical user interface [GUI] which includes access to the embedded MongoDB Shell, mongosh) and MongoDB Database Tools.

Launching MongoDB

To start MongoDB, first locate the file path of the MongoDB daemon (e.g., C:\Program Files\MongoDB\Server\5.0\bin\mongod.exe). Next, run the executable in Windows Command Prompt:

PS C:\Users\username> "C:\Program Files\MongoDB\Server\5.0\bin\mongod.exe"

Next, add the path to the mongosh.exe binary (i.e., the MongoDB Shell) to the PATH environment variable, or specify the full path at the command line. Launch the MongoDB Shell to connect to the MongoDB instance, where username represents the username of the Windows user:

PS C:\Users\username> "C:\Program Files\MongoDB\Server\5.0\bin\mongo.exe"

One can also connect to the MongoDB database via MongoDB Compass. Connect to a local database by passing an empty value as the connection string.

Accessing DataLENS Data

Please note that, due to restrictions imposed by the Data Use Agreements (DUAs) of the source institutions, the DataLENS database cannot be made publicly available. Access to a subset of the data may be provided upon request (e.g., to CS50 course staff).

Data Import

To import the DataLENS database, please complete the following steps:

1. First, show all existing MongoDB databases by running:

   show dbs

   An analogous command, show collections, can be used to identify all collections (i.e., MongoDB analog to a SQL table) in a database.

2. Switch the context to a non-existing datalens database:

   use datalens

3. Download the DataLENS database dump (in .bson format) to a local directory (e.g., C:/Users/username/Downloads/datalens/).

4. To restore the DataLENS database, run the following at the Command Prompt (i.e., not in the MongoDB Shell) at the directory where the database tools are installed (e.g., C:\Program Files\MongoDB\Tools\100\bin):

   mongorestore -d datalens "C:/Users/username/Downloads/datalens/"

5. If import was successful, running show dbs again in the MongoDB shell should reveal an additional database. Note that host name localhost and port 27017 are the default for local MongoDB connections.

Launching DataLENS

After completing the above steps, you are ready to launch Alzheimer DataLENS!

1. First, download the CS50 final project submission files, or clone the DataLENS GitHub repository (access available upon request).
2. Open the datalens R project in RStudio, then open ui.R, which defines the user interface, and server.R, which defines the back-end server logic.
3. Load all the libraries specified in server.R (i.e., run lines 8-30).
4. Click Run App in the upper right side of the RStudio editor. Alternatively, run shiny::runApp() in the R console. This will launch an instance of Alzheimer DataLENS as a locally hosted Shiny application.
5. To open DataLENS in a web browser, click Open in Browser or navigate to the local URL of the port where Shiny is listening on (e.g., http://127.0.0.1:3704).
6. Interrupt R to stop the application by pressing Ctrl + C, Esc, or the red STOP icon.

Deploying to Production

To deploy DataLENS in production mode, it is recommended to explore open-source Shiny Server hosted on an Amazon Web Services (AWS) Elastic Compute Cloud (EC2) instance with the cloud-native MongoDB Atlas service rather than a local installation of MongoDB.

Using DataLENS

To use Alzheimer DataLENS, please complete the following steps:

1. Summary information about DataLENS and AD is provided on the Home tab.

2. First, navigate to Input Genes and provide a list of genes to DataLENS. Then, validate the gene set using genome-wide human annotations to confirm that all identifiers are found. The following identifiers are accepted: HGNC symbol, Ensembl ID, ENTREZ ID, and UniProt ID. Descriptions and synonymous identifiers for validated genes will be displayed, along with a list of any invalid genes provided.

3. Next, navigate to Differential Expression Analysis. Select a gene of interest (from the list of genes validated in Input Genes), and examine the records of differential expression analyses pertaining to that gene. Note that logFC represents the log2 of the fold-change (i.e., effect size, often between case and control, also known as the M-value), P represents the nominal p-value (i.e., statistical significance), and adj P represents the p-value adjusted for multiple comparisons. The variable t represents the moderated t-statistic, or the ratio of the M-value to the standard error, while Ave. Expr. represents the average expression value for that gene. For additional information, please see the limma package documentation.

   Next, select various datasets in the dropdown menu. Observe how the expression levels of this gene change across various datasets. The bar chart will dynamically update to display both effect size (logFC) and significance (p-value, as color) across datasets of interest. Hover over each record to display full details in a tooltip.

4. In Interaction Network, select genes of interest. Investigate interactions between these genes in the cellular interactome, retrieved from the STRING database of known and predicted protein-protein interactions (PPIs). Nodes in the network are colored and scaled according to their differential expression [i.e., color is logFC while size is -log10(p-value)] in a dataset of choice. The weight of each edge represents the combined score (across gene neighborhood, gene fusion, phylogenetic profile, coexpression, experimental evidence, database records, and text mining) for interaction between two nodes. Note that the user must select Update Network before the graph will render.

5. In Regional Expression, explore transcriptomic datasets across brain regions. Select brain region(s) of choice in the dropdown menu, and examine records for transcriptomic datasets corresponding to that brain region. Brain regions are also visualized in brain segmentation plots constructed using the Desikan-Killany cortical atlas and automatic subcortical segmentation atlas.

Video Presentation

I showcase Alzheimer’s DataLENS in a brief video available on YouTube at: https://youtu.be/kuJpuaAQ-Lk.

About Alzheimer’s Disease

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder which impairs memory and cognition, and for which there is currently no effective treatment nor cure. The threat which AD poses to global public health is underscored by the following:

• More than 6 million Americans live with AD today; in the absence of research advances, this number is projected to rise to 13 million by 2050.

• 1 in 3 seniors will die with Alzheimer’s disease or a related dementia (ADRD), more than breast cancer and prostate cancer combined. Further, during the COVID-19 pandemic, isolation and neglect of the vulnerable elderly caused deaths from ADRD to rise by 16%.

• In 2021, ADRD will cost the U.S. economy $355 billion; by 2050, this cost could rise to $1.1 trillion. In addition, more than 11 million Americans provided 15.3 billion hours of unpaid care for people with ADRD in 2020 ― this labor is valued at nearly $257 billion. Importantly, these statistics fail to account for the emotional toll on families and caregivers.

Source: Alzheimer’s Disease Facts and Figures, Alzheimer’s Association 2021

Acknowledgements

Alzheimer DataLENS was created by Ayush Noori for CS50 at Harvard College. DataLENS is an initiative of the MIND Data Science Lab in the MassGeneral Institute for Neurodegenerative Disease (MIND) at Massachusetts General Hospital.

Unless otherwise indicated, all code submitted was authored by myself. The DataLENS database was provided by the MIND Data Science Lab. The DataLENS logo was adapted from a digital asset commercially licensed from Envato Elements.

I thank Dr. Sudeshna Das, Zhaozhi Li, Rongxin Liu, and Emily Merrill for their advice and support.

Selected References

1. Bihlmeyer, N. A. et al. Novel methods for integration and visualization of genomics and genetics data in Alzheimer’s disease. Alzheimers Dement 15, 788–798 (2019).

2. Noori, A., Mezlini, A. M., Hyman, B. T., Serrano-Pozo, A. & Das, S. Systematic review and meta-analysis of human transcriptomics reveals neuroinflammation, deficient energy metabolism, and proteostasis failure across neurodegeneration. Neurobiology of Disease 149, 105225 (2021).

3. Noori, A., Mezlini, A. M., Hyman, B. T., Serrano-Pozo, A. & Das, S. Differential gene expression data from the human central nervous system across Alzheimer’s disease, Lewy body diseases, and the amyotrophic lateral sclerosis and frontotemporal dementia spectrum. Data Brief 35, 106863 (2021).

4. Das, S., Li, Z., Noori, A., Hyman, B. T. & Serrano-Pozo, A. Meta-analysis of mouse transcriptomic studies supports a context-dependent astrocyte reaction in acute CNS injury versus neurodegeneration. Journal of Neuroinflammation 17, 227 (2020).