Experiences from the Data Science for Planetary and Human Health Symposium
While I attended this symposium on June 10-11, 2024 and I wanted to write this blog post earlier, I’ve been very busy and it was always low on the priority list. Hence why this post is written so late after the event.
I attended the Symposium: Data Science for Planetary and Human Health - The True Life Cycle of Multi-omics Data Conference 2024 on June 10 to 11, 2024 in Copenhagen, Denmark. This symposium was hosted by the Novo Nordisk Foundation Science Cluster as part of their efforts and activities to bring together different disciplines to share and discuss more technical work within data science that is being done in various fields of research. The summary they provide for the symposium on their homepage is:
“This conference aims to create a bridge between different disciplines and highlight versatile technologies, methods and algorithms presented by renowned researchers across academia and industry. The main topics encompass the full life cycle of research data including Data Management and Infrastructure, Data Analysis and Machine Learning, and Data Communication. The conference will provide a unique platform for the exchange of ideas, knowledge, and experiences in the field of FAIR data management and data science and will foster collaborations across disciplines.”
Since one of the main aims of the Seedcase Project is to develop software that simplifies doing research data engineering and building up data infrastructures, we felt that we fit in to the “data management and infrastructure” section that they list. We submitted an abstract and I gave a poster presentation about the Seedcase Project. This post shares some general impressions and experiences I had from attending this symposium. These are honest, if sometimes critical, reflections on what I saw and heard at the symposium (and often experience at other conferences too).
Issues common to most conferences
To start, many of the things I experienced are very typical of most other scientific conferences, so in that regard these experiences aren’t unique to this event:
Presentations contain too many graphs that are not explained enough. I see this so often in scientific settings, where a presenter shows graphs upon graphs of results, while quickly moving through them without explaining them at all. This style works well when the audience is intimately familiar with these types of graphs and the results that they show, but in a conference setting you can’t assume everyone will have that familiarity.
Presentations that make many assumptions about the knowledge of the audience. Some researchers (that I’ve personally talked to) who present their work often have a fear of looking like they do not know what they are talking about, so they overcompensate by presenting in a very detailed, jargon-heavy way. They assume the audience knows everything they are saying, but often this makes their presentation completely inaccessible to anyone outside of their narrow field of research. It often (at least in my experience) gets to the point where I don’t understand nearly anything in the presentations nor what their points are. It makes me feel like my time and presence aren’t respected, since I take time out of my work week to attend these events to learn and hear what others are doing. The least I hope for is that the presenters respect me and the audience enough to spend their own time to make sure the presentation is accessible and understandable.
Very little of what is presented is open (publicly accessible), even for projects that are more code- or methods-based. This is a general issue across nearly all of science. Collectively, we don’t share enough and we don’t make our work open and publicly accessible enough. This makes it really hard to assess the quality of what is being presented and makes it hard to know how to actually make use of the work that is being presented.
Unmet expectations
Considering the emphasis on the overall data science lifecycle, I had made a few assumptions about the symposium that were not met:
There were very few topics or sessions on data engineering, infrastructure, and management. I felt this was a bit of a lost opportunity for the symposium, since part of its stated aim was barely included in the program. It highlights to me, and reinforces my own personal experiences, that these topics are just not well known about, the awareness of them is nearly non-existent, and the value of them is very underappreciated. This is a major issue, since these are the foundational aspects of data science and research, and without them, the rest of the data science lifecycle is not possible.
A lot of presentations and topics were on data in general (or rather data that was collected) and on the analytics. This is fairly common in other conferences as well. This relates to the previous point: so often in research, we focus on the data collection and the analytics, but we don’t focus on the data engineering and infrastructure needed to support them. Even within grant applications, we write about what data we will collect and how we’ll analyze it, but we don’t write about how we will store, manage, organize, and share (even internally with the research group) that data.
Discrepancy between talking about FAIR principles and adhering to FAIR principles. While there were a few talks on FAIR principles, I didn’t see much evidence of them being followed in the presentations or in the projects themselves. This is a very common problem in science (at least in health research), where we talk about the importance of open science, FAIR principles, and reproducibility, but as researchers we don’t actually follow them.
Too much focus on genomics, and not other -omic fields. I was a bit surprised about this, as -omics is a huge field, but a large majority of the presentations and work were in genomics. I had been hoping to learn and hear more about other -omic fields (as I have some experience in lipidomic and metabolomic research), but there was very little of that.
Interesting insights and ideas
I did get some interesting insights and ideas from the symposium that I think might be useful to incorporate or consider in some way in the Seedcase Project:
There are a lot of -omic based projects, so we need to consider how those types of -omic data like genomics or metabolomics will be input into Seedcase software.
One presentation by Alexander Henriksen was on tracking energy use when running analyses on the server. It made me consider how much energy our work will use and what we could potentially do to lower it.
Further emphasize the strong fundamental need for more data engineering practices, even though the demand for these skills, knowledge, and tools hasn’t yet reached a conscious level in most researchers’ minds.
Even projects that are doing similar work as ours towards building data infrastructures are closed source or not publicly accessible and it isn’t obvious how they can be used in other contexts. This highlights the niche that Seedcase is aiming to fill in the research community (and also in small- to medium-sized companies).
Most projects and initiatives are not guided and driven by strong (or even explicit) principles, values, and designs. Building up these principles and designs take time, so it can seem that we are slow with producing software. But with these strong cultural and engineering foundations, we establish a base to do effective, rigorous, and high-quality work that we hope will last well into the future.
There were a few presentations I really appreciated and enjoyed, which also happened to be about how to give effective presentations:
The presentations by Eleonora Nigro and Anika Gupta on communication, specifically on accessibility via visuals and on using effective graphics were particularly interesting. This is especially important since they bring awareness to these issues that are often overlooked or ignored in scientific conferences and similar types of events. Coming from someone (me) who has difficulty differentiating colors, I really appreciated the emphasis on accessibility.
One of the final presentations by Lars Juhl Jensen was especially good, since he showed how to give an engaging and informative presentation. His style emphasized using simple slides, mostly just a few words or a single image, while keeping the entire focus on him and his speaking and presence. While I’ve heard some say (or fear?) that there isn’t enough “content/results”, the major advantage to this style is that it is more accessible to a wider audience of people from different fields of research or work, countries and proficiency in English, and in neurodiverse capabilities.
Summary
Overall, it was a decent conference. I had hoped for more topics on data engineering, more on other -omic fields, and on FAIR practices, but unfortunately I was a bit disappointed. I did get some interesting insights and ideas from the symposium that I think might be useful to incorporate or consider in some way in the Seedcase Project. It also further reinforced the importance of the work we are doing with Seedcase.