Welcoming ToPCaP’s newest working group: Computational Pathoepidemiology!
January 28, 2016

Welcoming ToPCaP’s newest working group: Computational Pathoepidemiology!

Sparked by previous ToPCaP discussions, the Computational Pathoepidemiology group, led by Dr. Travis Gerke and Dr. Svitlana Tyekucheva, made its debut in late 2015. The group strives to be an inclusive network, connecting scientists who are interested in talking about the computational aspects of prostate cancer research. During the group’s monthly meetings, members brainstorm collaborations for new projects, discuss different directions to go with new data, and evaluate the effectiveness of computational tools and methods.

Computational Pathoepidemiology holds exciting promise, and we interviewed Dr. Gerke and Dr. Tyekucheva to learn more about the two of them and how they see the group: its culture, its goals, and its place within ToPCaP. Dr. Tyekucheva, whose favorite fictional character is Agatha Christie’s legendary detective Hercule Poirot, is a bioinformaticist. When he’s not reading Catcher in the Rye, Dr. Gerke is busy eating hot wings and doing science as a biostatistician/epidemiologist. Their favorite computational methods are not flashy: they believe in the power of simple linear regressions and the t-test to explain complex phenomena.

The following is everything you wanted to know about the Computational Pathoepidemiology group!

Q&A with Travis Gerke and Svitlana Tyekucheva, interviewed by Alex Tinianow after a Computational Pathoepidemiology meeting

Alex: Thank you two for joining me for this interview! Here’s the first question: Why make this group? What’s the point?

Svitlana: Well, because it’s fun! It’s nice to have a group of people who are interested in the computational aspects. And it promotes rigorous computational approaches to our studies.

Travis: I think the composition of the group is healthy for research. We have Jen Sinnott, who is a statistician ninja; Svitlana, who is a bioinformaticist master; Lorelei Mucci, who is our big picture visionary; and then many more fantastic people in between. It’s a good mix, I think, compared to just having a bunch of pure epidemiologists or pure bioinformaticists or computational statisticians.

Svitlana: We also wanted to have something that was a smaller group from the big group, and there was a need for a computational workgroup where we can discuss challenges that we face. So it’s not a forum for polished presentations -- it’s for discussions and brainstorming what can we do with the data? Because now, we have a lot of data collected, and this is a good place to discuss what we can do with the data next. Both the data science and prostate cancer research questions are there, and then together we can figure out what we can actually answer given what new computational tools we have. And there was a need for that, a brainstorming time slot.

Travis: I think that’s very true, because in many of our team meetings, conversations ultimately lead one of us to ask “what is the general tool that we should be using to answer this question?” Of course, the underlying machinery might not be of interest to everyone, and that’s totally fine. The nature of this smaller group is to help refine the way we meet and talk about methodological issues: how to find the tools and even new data sources. As Svitlana said, when we stumble on a new data resource, we want to talk about it but don’t have polished results to discuss or anything like that, so it might not fit into other working groups. But in comp pathoepi we can just say “hey, there’s this thing, what should we do with it?” I think that’s useful, it makes it fun.

Alex: So how does it relate to ToPCaP?

Travis: I think it fits into the context of ToPCaP very well, because in the ToPCaP group, so many of the projects are molecular studies, high dimensional, big data projects, and in that setting there will inevitably be overlap with the methods and tools we talk about in the computational pathoepi group. So I think whether or not we’re involved directly yet, we would love to be!

Svitlana: Yeah, and then people, if they have challenges, have a forum to discuss them. And even if they’re not regularly at the meetings, but they need some advice and they don’t have, or they don’t know, people with the necessary computational expertise at their institutions, they can always talk to us. We’re glad to talk if somebody has questions that can be directed at us, so we can help out with their computational aspects.

Travis: It’d be good if we had more visiting people attend our group, with the project and then the need. We’d love for more people to come, even if they’re not a computational pathoepidemiologist per se. If they just present their idea, then we can all talk about how it’s fun and interesting, and then collaborate and move it forward. We always like new things to talk about, and it would help to have more examples to draw on, just for discussion about tools that work or don’t work.

Svitlana: And also, even people who are interested in listening to the type of things that we discuss, for educational purposes, could sit in. Other groups, they may have their own more computational folks - it’s not that only people who just need some help with their analysis should come, but also people who run analysis in those groups. We could share our expertise and talk about packages we use, and we can broaden what we know.

Alex: Do you have any specific hopes for the impact of this group?

Svitlana: That it will be useful, that we can make a difference, and that we can really bring together scientific questions, computational resources, and our computational abilities. And that we can, in the process, come up with exciting projects, ideas, and gadget-y kind of tools that everyone can use and enjoy. That will move us forward and also keep it fun. And, of course, the computational part should not be a bottleneck to answering the questions that we have.

Travis: I think streamlining everything that we do is a goal. I mean not only be able to write papers that are fun, but also that we can write quickly with the right tools in place. Once we have a solid collection of tools that we all agree that we like and work well for a variety of purposes, it becomes easy to answer important questions, instead of just wishing that we knew how to write the right script or find the right tool. That’s all ancillary stuff to the ultimate dream of the broader pathoepidemiology group. Although that ancillary stuff is what the computational pathoepidemiologists find fun!

Alex: So then is there a certain project you’re really excited about working on?

Svitlana: I’m excited about what we’re still working on: the tumor stroma project. There’s such a wealth of data so there’s still more to be done, so this is interesting. I’m also very excited about the SECRET MYSTERY R01 grant that the US group will be submitting soon. It’s phenomenal, because if we get this data, it’s not even just answering the questions that are scientific prostate cancer related questions – it’s more of creating a resource, and then we can try to answer whatever we want with that. This is so amazing.

Travis: Yeah I agree. You can call it “Secret Mystery R01 that we’re submitting in February.”

Alex: Alright, I’ll keep it secret. Thanks so much for your time and for telling us all about this exciting new group!