Plan your data management

Planning a research project can be at times overwhelming especially if it is our first time approaching such a task. While designing a project we need to consider a number of technical aspects concerning the data we would like to produce, however we often underestimate how to manage efficiently our data. Thinking ahead about the best way to manage our data can save us plenty of time in the future and drastically reduce our mistakes.

A good starting point to think about data management is to create a Data Management Plan.

What is a Data Management Plan (DMP)?

A data management plan is a research document, tailored to your project, that, as the name suggests, helps you to organize your current and future data. It simply consists of a questionnaire that guides you through some of the aspects of the data life cycle.

Despite its name, a DMP should be considered as a living document that you can update throughout your project. Keeping your DMP up to date allows you to record essential information relative to your data, such as sparse information and metadata that you might need in a later stage of the project (e.g. when writing a paper or a report for a funding agency).

A DMP is increasingly required by funding agencies, since 2021 it is indeed mandatory to provide a DMP for all the European grants, here you can find a DMP template by the ERC.

How to create a DMP

There are many options to create a DMP, you could write your DMP by following templates from funding agencies, or you could use ad hoc online tools such as:

• Argos
• dmponline
• Data stewardship wizard
• rdmo project

Why should you use a DMP?

As previously mentioned, a DMP is incresingly becoming mandatory to acquire fundings from German and European agencies, but besides this incentive it provides some other real advantages:

• It helps you to assess critically different aspects of your data that you might easily overlook.
• It helps you to think of your data organically.
• It forces you to create documentation and proper metadata at an early stage of your project.
• If properly updated during the project life, you can publish the DMP together with your data.

Start writing documentation right now

Documenting your data only at the end of your project is not advisable. It forces you to rely on your fallible memory, on sparse information and it is extremely error prone and time consuming

Planning your project at Ipsy

In order to help you manage your data, we have implemented a basic research project workflow that requires very little effort from your side.

To adopt the project workflow we suggest you to use the cluster for data analysis and data storage. The project workflow consists of three simple steps:

1. Fill in a brief questionnaire that helps us to create some intial metadata for your project. Further information about the questionnaire can be found here.

2. Once the questionnaire is properly filled in, we create a dedicated project which comes with a predefined folder structure. In order to have detailed information about the structure, please refer to the specific project structure section.

3. Once your raw unprocessed data have been moved into the predefined structure, your data need to be converted into the BIDS format (see BIDS) by using the tools available in the cluster.

Data Version Control

One of the most important aspects of data management is to keep track of the evolution of your data. This task might be particularly error prone if done manually, it is, thus, advisable to use a systematic and automated approach.
As for software development also data can be tracked via version control. There are not many version control tools dedicated to data, the main one is Datalad, which in the last years has been heavily adopted in cognitive neuroscience research.
Datalad does not simply track the changes of your data, but also records their relationships. Beyond that, the datalad ecosystem includes other tools that can help you to make your science reproducible and collaborative.

The best way to start with Datalad is to go through the Datala-handbook. Dalalad is natively used through the command line, like git, but for those who are less comfortable with the terminal there is also a GUI version.