Research data lifecycle diagram showing plan, collect, process, analyse, preserve, share and reuse
English

Research Data Management: understanding the data lifecycle in real research

TL;DR (30 seconds)

  • Research Data Management (RDM) is about managing research data from the beginning of a project until long-term reuse.
  • It is not just storage. It includes documentation, quality, privacy, preservation, and sharing.
  • The RDMkit lifecycle model (ELIXIR) helps structure research work into clear phases.
  • Good RDM improves reproducibility, transparency, and research impact.

Before starting: why does RDM even matter?

During the EP PerMed training at ELIXIR, I realised something important. In research, we usually focus on results. But results are only as strong as the data behind them.

Many research problems happen because:

  • data is poorly documented
  • folders are messy
  • no one remembers how preprocessing was done
  • sensitive data is not handled correctly
  • results cannot be reproduced later

Research Data Management exists to reduce these risks.

What is Research Data Management?

Research Data Management (RDM) covers everything related to how research data is:

  • planned
  • collected
  • processed
  • analysed
  • preserved
  • shared
  • reused

It is not a single task at the end of the project. It is something that should be considered from day one.

RDM connects documentation, metadata, ethics, GDPR compliance, reproducibility, and open science.

The Research Data Lifecycle (RDMkit Model)

During the training, we used the RDMkit data lifecycle model from ELIXIR.

Research Data Lifecycle according to RDMkit

This model shows that research data moves through connected phases. It is not strictly linear. Reuse can start a new cycle.

The full framework is available here:

👉 https://rdmkit.elixir-europe.org/data_life_cycle

The lifecycle stages explained in simple terms

1) Plan

Before collecting any data, it is important to decide:

  • What type of data will be generated?
  • Which formats will be used?
  • How will data be stored?
  • How will privacy and consent be handled?
  • Who will have access?

At this stage, a Data Management Plan (DMP) is often written. It should not be static. It should evolve during the project.

2) Collect

Data collection includes experiments, surveys, simulations, or observations. These are important aspects:

  • Good experimental design
  • Data quality (consistency, accuracy, completeness)
  • Recording problems during collection
  • Using electronic systems like ELN or LIMS

Metadata should be captured at this stage, not added months later.

3) Process

Raw data is rarely ready for analysis. This stage includes:

  • Cleaning
  • Quality control
  • Transformations
  • Integration of multiple datasets
  • Early anonymisation for sensitive data

Every transformation step should be documented clearly.

4) Analyse

This is where insights are generated. Good practices may include:

  • Documenting workflows
  • Saving parameter choices
  • Commenting code
  • Maintaining version control
  • Keeping a clear folder structure

Reproducibility depends heavily on this stage being well documented.

5) Preserve

Preservation ensures long-term usability. This includes:

  • Using appropriate file formats
  • Clear file naming and versioning
  • Writing README files
  • Backups and snapshots
  • Using checksums to detect corruption

Without proper preservation, data can become unusable even if it still exists.

6) Share

Sharing data increases transparency and citation potential. The Key decisions are:

  • Which repository to use
  • What license to apply
  • What metadata to include
  • Whether access should be open or controlled

Sensitive data does not need to be fully public. It can follow controlled access rules.

7) Reuse

Reuse allows data to:

  • Validate results
  • Support new studies
  • Be combined with other datasets

For reuse to work, data should follow the FAIR principles:

  • Findable
  • Accessible
  • Interoperable
  • Reusable

RDM and Open Science

RDM supports open science by:

  • Improving reproducibility
  • Reducing publication bias
  • Increasing transparency
  • Making data reusable beyond the original project

Good data management benefits both individual researchers and the broader scientific community.

Personal reflection

As someone doing a Master’s in Artificial Intelligence and coming from a software engineering background, I found the lifecycle approach really practical.

In software projects, we naturally think about structure, version control, documentation, and system design from the beginning. In research, however, data is often treated as something temporary, just a step toward publication.

What I learned during this training is that data should be treated like a product, not a by-product.

The lifecycle model helped me understand that every stage — from planning to reuse — affects reproducibility and long-term value. Small decisions like naming files clearly, writing metadata properly, or documenting preprocessing steps can make a huge difference months later.

Planning early saves a lot of effort later.

Resources

If you want to explore Research Data Management and the data lifecycle in more detail, these are useful references:

These resources are especially relevant for European research projects and for researchers working with sensitive or complex datasets.

Conclusion

Research Data Management is not just a requirement. It is a way to make research more reliable and sustainable.

The data lifecycle shows that reproducibility and reuse depend on decisions made from the very beginning of a project.

Treating data as a valuable research output improves both scientific quality and long-term impact.

Ayesha Munir — Software Engineer | Artificial Intelligence (MSc)