The DIAL Workbench is a set of tools and recommendations how to conduct research in our lab, with focus on statistical analysis, reference managing, version control, and manuscript preparation. We update the Workbench every year and try to optimize our processes. Most of the DIAL Workbench is mandatory for lab members to allow seemless team work. If you just want to see and test how we work, this page will serve as a starting point.
The DIAL Workbench 2018 includes:
Mendeley is a free reference manager from Elsevier, with which you can organize and cite all your literature. Mendeley integrates nicely in Word, and library entries can be directly created from Firefox or Chrome. We like Mendeley for its ease of use, its good functionality when working in teams, its ability to automatically complete entries by querying DOI and pubmed, and its iOS app, which lets you read and annotate PDFs on the go. You need to create an online account, and your library is additionally stored in the cloud, which greatly simplifies team collaborations, but requires some caution when referencing unpublished work (which you can flag as such).
Alternatively or additionally we recommend the use of Zotero. Zotero is very similar to use and shares many functionalities with Mendeley. Each has some strengths and weaknesses, but ulitmately it comes down to personal preference. You can also use both; Mendeley nicely synchronizes with Zotero, however, files like PDFs are not synchronized. We like Zotero over Mendeley for being an academic project, its good organisational features like virtual folders based on queries and tags, and its fast and lean user interface. Team collaborations are a little bit less convenient than with Mendeley, and it has no native iOS app (but 3rd party apps support it).
The R Project for Statistical Computing
R is a free software environment for statistical computing and graphics. R is incredible powerful and is our weapon of choice for solving 99.9% of our statistical analysis problems. The learning curve especially for non tech-savvy people is somewhat steep, but you will quickly reach a level where R will outperform most other statistical software. Please be aware that R is updated regularly. While there is no clear guideline, when and how you should update, we recommend using the latest version, which we just assume had some bugs fixed.
RStudio is a very nice and user-friendly user interface for R, and we recommend using it for your everyday work with R. While we tested many other GUIs, RStudio consistently came out as the best choice. It nicely integrates with many of the other DIAL Workbench tools, including git. And it can do tex.
R is extended by packages, and you will need many different packages in your daily work dependent on your project. One package that is absolutely necessary for the DIAL Workbench to work is rmarkdown. It will take your statistical analysis to the next level by allowing automatic generation of HTML, Word or other reports. If you want, you can write your entire paper in rmarkdown and then create it on the fly (but we recommend this only for geeks).
You can install the rmarkdown package by typing in R:
Another R package we recommend for the core set is tidyverse. Tidyverse is a collection of several R packages that evolve around data management and graphics creation. Most importantly, it includes ggplot2, which we use for most of our graphics needs.
You can install the tidyverse package by typing in R:
Version control allows us to backup all versions of an evolving analysis script or rmarkdown. This makes tracking changes and finding potential errors as well as collaboration much easier. git - like R - works in the background and is needed by the below listed frontends as a backend software.
We use GitHub as our cloud repository for all R and other analytics code. While usually allowing private projects only in paid programs, as an academic/educational institution you can request exemption. As a non-profit academic research institute we are allowed to host unlimited private projects within their education program. You can register for the educational program here: https://education.github.com/. Please contact Alexander Brandt if you need access to our lab's repository.
While git/GitHub works integrated in RStudio, Slack, and other relevant software, we recommend additionally using a desktop application, which allows direct file system operations on git repositories. There are many applications available, but for the DIAL Workbench we prefer GitHub Desktop.
For lab notes and project coordination we use a labfolder installation, which is provided by the Berlin Institute of Health. Labfolder completely replaces a written lab notebook with all legal requirements from GCP. Please contact Alexander Brandt, if you need access to our internal labfolder installation. It can only accessed via Charité intranet or via VPN (you need a special service extension for regular VPN).
Following an open access policy for publishing raw data from research papers we encourage all colleagues to publish raw data regarding their publications on OSF. Data should be carefully curated to not contain any personal information. For this reason, all uploads to OSF (preferably directly from GitHub) need prior approval by the respective team leaders. Further, the OSF repository should be referenced in the acknowledgements of the respective publication.
For tex documents we highly recommend Overleaf as an online collaborative tex editor. Overleaf tex files can be cloned to github, which makes both online and offline editing with a different tex editor of your choice possible. Please contact Amir Motamedi how to connect github and overleaf.
For electronic data capture we use a REDCap system, which is hosted by the Berlin Institute of Health. We capture all data in our investigator initiated clinical studies using REDCap. And recommend using REDCap as the main database and data entry point for all studies and projects, even small ones. Please contact Maja Olszewska, if you need access to our internal REDCap installation or to specific projects. Our REDCap system can only be accessed via Charité intranet or via VPN (you need a special service extension for regular VPN).
We use a self-hosted XNAT system as our image repository. Please contact Michael Scheel, if you need access to our internal XNAT installation. Our XNAT system can only be accessed via Charité intranet or via VPN (you need a special service extension for regular VPN).