Diversity in Datasets

For all people to truly benefit from scientific advancement and the full potential of genomics, diverse datasets must be used for research and clinical care. But at all stages of genomic research, there is a critical lack of dataset diversity. The GA4GH policy tool provides research teams with a structured, actionable framework to define and pursue diversity in human genomics and genetic studies by emphasising that diversity is a means, not an end. It asks teams to clarify what types of diversity are relevant (i.e. ancestry, geography, socio-economic status, health status, gender, environment) and why they matter for their specific research goals. Then, it guides them on how to embed those diversity considerations across the full data lifecycle – from study design, participant recruitment, data collection, analysis, sharing, and interpretation. The framework also highlights the need to engage with communities, address barriers to inclusion, avoid conflating categories (e.g. genetic ancestry versus ethnicity), and ensure that diversity strategies align with normative values such as justice, beneficence, and respect for persons. By doing so, the tool shifts genomics studies from simply striving for representativeness toward making deliberate, context-sensitive design choices that maximise the scientific and equity-oriented value of diversity.

There is well known bias in current genomic datasets with much of the data being from individuals of caucasian ancestry. In the current ecosystem, there is a critical lack of dataset diversity across the pipeline of genomic research. In order to optimise the benefits from advancements in genomic research, research infrastructures must strive to become equitable and inclusive by considering more diverse data.  Developed by the GA4GH Regulatory & Ethics Work Stream (REWS),  the Diversity in Datasets policy explores types of diversity  and shares actionable recommendations for researchers in order to uphold diversity in their research and findings.