AGENDA

REWS Diversity in Datasets Meeting 2
29th June 2022 @ 11am UTC/12pm BST/7am ET/1pm CET/9pm AEST/7pm Hong Kong

(dial in numbers below)

[Zoom Recording Passcode: Gu^5%eb.].

Zoom Details:

Join Zoom Meeting

https://us02web.zoom.us/j/84490365385?pwd=MmljSUtjSzV1OTZybnBaL0ZORUc4QT09

Meeting ID: 844 9036 5385

Passcode: 036226

Attendees:

1. Olivia Ngan
2. Maili Raven-Adams
3. Anna Lewis
4. Ariana Gonzalez
5. Bimal Chaudhari
6. Hanshi Liu
7. Aime Lumaka
8. Ainsley Newson
9. Sasha Henriques
10. Maxine Mackintosh
11. Ashley Hobb
12. Lindsay Smith
13. Thijs Devriendt

Minutes:

AL: Human genomics context for use - a lot of use on the discovery side for establishing genotype-phenotype links, gaining a mechanistic understanding of genetic variation and understanding population history. On the applications side there’s genetic testing and development of interventions (using genetic knowledge to develop drugs or structural interventions).

Diverse Data x GA4GH

MM: GEL has gone through a transition in the past years, now that the 100k is done. Now a platform that provides genomic sequencing services to the NHS and make it available for research. Closing the infinity loop. Newborn screening, cancer 2.0 and diverse data programs. Diverse data is focused on all patients regardless of background receiving the same level of care. Reduce health inequalities to achieve trust, better research and improvements to diagnosis, treatment. 4 work streams - understand the data gap, community and engagement, filling the gaps with sequencing, and product, tools and behaviors (eg. R package to a handbook on how lead certain discussions).

We did a literature review into definitions of diversity in genomics - key highlights include inconsistencies in how ancestry is defined, absence of intersectional definitions of diversity, socioeconomic groups are vastly underrepresented, no one seems to articulate why a sample distribution was selected or how it was accounted for.

The outcome you’re looking to optimize for really defines what your diversity means, real spread of definitions. Took these outcomes and compared them to established bioethics principles and created an evaluation metric of all the ways to think about what underrepresentation means. It’s all about values and how you are embedding that in your study design. Expanding into a toolkit.

Another project to flag is lost in translation, a toolkit to provide guidance on what terms to use in what way. Force precision in language which influences decisions.

Created a metaphor for working through biases and implications of decisions.

About to launch a genomic data diversity initiative, the world's biggest open source library of curated tools and products. Wonderful to have as many people involved as possible, welcoming people to add models, tools, etc.

Discussion

AL: Looking for definitions of diversity project, saw you were points to other definitions like diversity. Were people actually defining diversity?

MM: Yes people did define it, but the default is thinking about ancestry.

AL: There are these multiple different definitions that are important. Wondering about a framework sitting on top of that that defines which dimensions are important.

MM: We landed on values being the thing that drive that. Just the process of encouraging people to be thoughtful about the outcomes has made a big difference. Doing a proper factor or root cause analysis causes you to think about other things driving the differences.

MRA: Doing that whole process helps with transparency and explain why people have used a certain set of data or how they interpreted it.

MM: Underrepresented is vastly different depending on the context. We looked at the big ancestral problem, which we still need to address, but need to look at your own system and pipelines because it could paint a different picture.

AL: Brainstorming about what makes genomics special when we’re talking about genomics data. Importance of genetic ancestry is a clear headliner, but it’s not the only one. Is genetics special?

BC: I don’t think it’s as special as we think it is, I think genetic exceptionalism that is dangerous. Majority of the problems of diversity or bias we’re just re-inventing a number of socially based ills in healthcare. One thing I have seen that is uniquely genomic is the potential to use genomics and genetics as a way to put a scientific definition on racism. The biggest value add is to highlight all of the ways we’re not special, and borrow from our colleagues.

AN: Broadly agree with that. Genetic exceptionalism literature has been around for 20 years. Would encourage people to take a look at it. Other thing is that this depends on what we’r talking about because genomics is not a single entity. Think about use cases or questions where this is particularly important.

AL: Framed as genetics different to healthcare, but there is also the research side.

ALu: I think genetics is particular in terms of daily challenges in rare disease and complex disease.

AL: The daily challenges, are you talking about more uncertainty than other fields?

ALu: For example, genetic sequencing on someone with African descent who has a number of VUSs. How to approach this? Don’t have the resources readily available. Expanding on the meaning of diversity - depending on what you are looking for, diversity is something different [shared screen]. We see diversity in two terms - geographical capture and the smaller groups.

HL: Fairly certain that there is ther perception of genetic exceptionalism in the general population and in some organizations. I believe there are some countries with laws in genetic exceptionalism. Whenever you make scientific discoveries, these prevailing ideas make scientific communication dangerous in a way if you don’t do it carefully. Challenge in how we communicate about it. No matter how much we dismiss it, it’s going to be something that’s going to catch up.

SH: The fact that physicians and publics think that there is genetic exceptionalism is was gives it the power. Generally, there isn’t an area of healthcare where genetics doesn’t have a part. Big interdisciplinary way with lots of medical science and technology.

AN [from chat]: That is so interesting Sasha - empirical genetic exceptionalism even if not conceptually justified or universal

MRA [from chat]: we can think about genomics independently, without ignoring all the other important factors at play

AG: When you write papers in argentina, have to be very careful with some terms for ethnicity that involve so much more than genetics. It’s not only the genetics but all the history you’re bringing behind it. In communications, words can get dismissed. Learn when and how to use certain words.

Next Steps:

 Next meeting to talk about international data sharing culture and the technologies and cultures that set it apart. Link onto what we want to achieve in the policy.

Will try to schedule a July meeting.

Will be circulating a Google Doc for comment and additions.