This is a Shiny-App with the aim of facilitating association and covariate selection for targeted metabolomics data from multiple studies with a wide range of factors (demographic, lifestyle, etc.). A detailed description of the implemented methods can be found here. Features of the App are:

A detailed description of the methods used in the analysis as well as an application with data for 63 metabolites and 29 factors in three studies (N=16,222) is integrated in the LHA (Leipzig Health Atlas) here.

• Upload of data (with seperate files for metabolite and factor data)
• Matching via ID overlap
• Preprocessing of data in each cohort including
  • Removal of outliers defined as values +5 times the sample standard deviation (5*SD) on log-transformed data
  • Inverse-normal-transformation of data to remove skew while regainig data structure (concerning zero-inflated values)
  • Nonparametric batch-adjustment via an empirical Bayes method (sva::ComBat)
• Univariable association step of each metabolite with each factor in each cohort (Metabolite_i ~ Factor_j)
• Univariable test for interaction of each factor with the cohort term (Metabolite_i ~ Factor_j + cohort + Factor_j*cohort)
• Correlation check and user guided descision on exclusion of too highly correlating factors
• Multivariable association step of each metabolite with all factors in each cohort (Metabolite_i ~ Factor_1 + Factor_2 + ... Factor_J)
• Multivariable test for interaction of each factor with the cohort term (Metabolite_i ~ Factor_1 + Factor_2 + ... Factor_J + Factor_1*cohort + ... + Factor_J*cohort + cohort)
• Selection of covariates for subsequent analyses of metabolites by removing factors not explaining a set amount of variance in at least one metabolite in at least one cohort via backwards selection until only factors meeting the explained variance criteria remain in the model
• Visualization of results
• Sample and feature-centric annotation
• Download of results

The App is hosted on the LHA-Servers here. Otherwise, it can be started directly from this repository via:

# to install shiny run:
# install.packages("shiny")
# load the Shiny package
library("shiny")

# Start the App directly through Github
runGitHub("Metabolite-Investigator", "cfbeuchel")

For local offline use the the application needs to be downloaded, the ZIP-file extracted an run like this:

# to install shiny run:
# install.packages("shiny")
# load the Shiny package
library("shiny")

# start the App from the previously downloaded folder, e.g.
runApp("C:\Users\cfbeuchel\Downloads\Metabolite-Investigator-master")

# or on unix based systems for example:
runApp("/home/cfbeuchel/Downloads/Metabolite-Investigator-master")

Replace the directory within the brackets with location of the folder on your system. Make sure the path points to the folder containing the app.R, server.R and ui.R scripts.

The App comes with example data from two cohorts that may be used to try out the functionality. Press the Use Example Data button to load the data and try out the application.

For backwards-compatibility, we also offer a docker archive of previous versions of the app on GitLab (starting from Version 0.1.6). These docker-images can be pulled from the archive and run within a working docker environmnent as follows:

1. Go to https://gitlab.com/imise-genstat/metabolite-investigator-archive/container_registry/1612096
2. Look for a tagged version you would like to run using docker
3. The download link for the docker command is available at the "clipboard"-symbol next to the tag-name (e.g. 0.1.6)
4. Use the docker pull command to download the selected image to your computer
5. Run a local version of the downloaded image

# Using the initial release version 0.1.6 as an example: 
docker pull registry.gitlab.com/imise-genstat/metabolite-investigator-archive:0.1.6
docker run -it -p 3838:3838 registry.gitlab.com/imise-genstat/metabolite-investigator-archive:0.1.6

The app requires an up-to-date R installation as well as the following packages:

• shiny (install via CRAN - install.packages("shiny"))
• DT (install via CRAN - install.packages("DT"))
• BiocManager (install via CRAN - install.packages("BiocManager"))
• sva (install via Bioconductor)
• data.table (install via CRAN - install.packages("data.table"))
• visNetwork (install via CRAN - install.packages("visNetwork"))
• magrittr (install via CRAN - install.packages("magrittr"))
• ggplot2 (install via CRAN - install.packages("ggplot2"))
• corrplot (install via CRAN - install.packages("corrplot"))
• scales (install via CRAN - install.packages("scales"))

This App could only be programmed by using these freely available tools:

• Martin Morgan (2018). BiocManager: Access the Bioconductor Project Package Repository. R package version 1.30.4. https://CRAN.R-project.org/package=BiocManager
• Yihui Xie, Joe Cheng and Xianying Tan (2019). DT: A Wrapper of the JavaScript Library 'DataTables'. R package version 0.11. https://CRAN.R-project.org/package=DT
• Jeffrey T. Leek, W. Evan Johnson, Hilary S. Parker, Elana J. Fertig, Andrew E. Jaffe, John D. Storey, Yuqing Zhang and Leonardo Collado Torres (2019). sva: Surrogate Variable Analysis. R package version 3.30.1.
• Matt Dowle and Arun Srinivasan (2019). data.table: Extension of data.frame. R package version 1.12.2. https://CRAN.R-project.org/package=data.table
• Almende B.V., Benoit Thieurmel and Titouan Robert (2019). visNetwork: Network Visualization using 'vis.js' Library. R package version 2.0.6. https://CRAN.R-project.org/package=visNetwork
• Stefan Milton Bache and Hadley Wickham (2014). magrittr: A Forward-Pipe Operator for R. R package version 1.5. https://CRAN.R-project.org/package=magrittr
• H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2016.
• Taiyun Wei and Viliam Simko (2017). R package "corrplot": Visualization of a Correlation Matrix (Version 0.84). Available from https://github.com/taiyun/corrplot
• Hadley Wickham (2018). scales: Scale Functions for Visualization. R package version 1.0.0. https://CRAN.R-project.org/package=scales
• R Core Team (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
• Winston Chang, Joe Cheng, JJ Allaire, Yihui Xie and Jonathan McPherson (2019). shiny: Web Application Framework for R. R package version 1.3.0. https://CRAN.R-project.org/package=shiny
• Achim Zeileis, Torsten Hothorn (2002). Diagnostic Checking in Regression Relationships. R News 2(3), 7-10. URL https://CRAN.R-project.org/doc/Rnews/