On the line: 13 questions for Thomas Hrabe, serial research tool developer

Today’s On the line is with Thomas Hrabe, a talented postdoc from the Godzik lab at the SBP Medical Discovery Institute, La Jolla, with a special interest in the structural biology of the nano machines that run this whole operation we call life, i.e proteins.

Thomas has developed several important tools in the field of molecular structure analysis, ranging from trendsetting electron tomography processing with PyTom, to scientifically insightful tools such as PDBFlex. Additionally, Thomas has contributed to the development of Cancer3D, a tool to track drug sensitive and driver mutations, and which he may soon proudly say made the transition from the bench to the bedside.

But let’s not give away any more spoilers, let’s hear it directly from the source. Thomas, take it away!

Tell us a little bit about yourself and the scientific interests you pursue.

I studied computer science at the Technical University in Munich and later got my Ph.D. from there, even though I technically worked at the Max Planck Institute for Biochemistry. I have always been fascinated by developing software for interesting data. For instance, during my CS studies, I was reading the rather broad journal “Spektrum der Wissenschaft” that publishes articles from all fields. The reason why I eventually ended up developing for Structural Biology (pytom.org) was a coincidence, but I was fascinated to learn about the tiny machines that everyone is comprised of. I am 100% computer scientist, but I am really intrigued by the biology we get to analyze, which is tremendously increasing in data size, resolution and accuracy, and of its importance for us humans.

What led you to develop webserver applications for PDBflex, Cancer3D, Pytom, and POSA?

I believe the web is a great tool to bring (any) processing to the right user. In our case, we are talking about scientists, and one problem I see with methods and scientific software is that it’s kind of “fire and forget”: publish and move on.

“I am 100% computer scientist, but I am really intrigued by the biology we get to analyze”

Few people really care about product maintenance if you write a pure terminal based application. Webservers allow non-experts to access your method without having to download and install them. It also forces you to make it really useable. It’s more of a challenge, but eventually more rewarding if it’s really accessible and used by a large audience.

Can you give us a quick behind-the-scenes tour of the Cancer3D engine and it’s advantages over similar existing tools?

Eduard Porta Pardo is the brains behind these methods, while I was the computational muscle to put the server together. However, the most distinguishing feature of the algorithms is the “resolution” (as we call it) with which the data is analyzed. We look for mutation statistics in protein domains and on interaction interfaces rather than analyzing whole genes. We found interesting correlations in patient survival rates that would otherwise be hidden by background noise.

Cancer3D is a fantastic tool. Do you think it can be applied to a clinical context sometime in the future?

The results can be used to assist MDs in the clinic, but many must be validated because those are new findings. One of our colleagues is looking for ways of turning this into commercial software. It’s really exciting, because we get to see, live, the process of technology being translated into usable methods.

Alright, this is a hard one… if you had to pick one of the tools which you developed, that in your opinion has the most impact, which would it be?

Ok, I have to choose two:

For its uniqueness, I choose localize.pytom.org – it’s the first ever webserver in the field. I am curious if the field will adapt to this concept and follow these footsteps. It’s standard in Bioinformatics, but not so in the field of electron microscopy (yet?). The internet bandwidth is available and should permit this transition to webservers.

From a scientific potential standpoint, it would have to be PDBFlex.

What tools were indispensible for creating PDBflex, Cancer3D, localize.pytom, and POSA?

Git version tracking for the most. All are basic LAMP design servers, there’s not much magic. However, in our current project, we are moving on to Angular because we have the freedom (time) to plan it out more.

Thomas’ Toolbox:

What tools do you use in your daily routine at the lab?

I think in C, but I mainly write Python and Javascript these days. Sublime editor is the best, Eclipse is great for larger projects, MacPorts for the tools to run on my desktop.

What important guidelines do you follow and do you have any important tips you might have for efficient resource development?

KISS (keep it simple stupid).

In your opinion, what is the most relevant programing language for the beginner bioinformatician/biologist?

I think many would answer Perl here for bioinformaticians, but do you really want to learn a language from the IT bronze era? Hence, I would not recommend it.

“It’s more rewarding if [your tool] is really accessible and used by a large audience”

In my opinion, as a computer scientist you are given a toolbox of programming languages that you must select wisely to fit a specific problem. But to answer the question: Python in the very beginning, R for statistical analysis and finally HTML / JavaScript if you want to do programming and visualization. Knowing JS is valuable for the future, especially if the wet lab does not work out as it should.

What role do you perceive collaborations between computational and wet bench biologists should have in life science research? Do you have any ideas of what an optimal framework for operating such a collaboration could look like?

Nothing beats a well balanced team where everyone does what they are trained to do and learns from each other.

“We found interesting correlations in patient survival rates that would otherwise be hidden”

A nice example for a failed project of someone attempting to work outside his own field and trying to get top results is in 
“Surely you’re joking Mr. Feynman” where a physicist talks about the outcome of him doing some ribosome purification and the consequences it had. An army of computer scientists will quite unlikely solve biological problems and vice versa, it takes a mix of expertise.

What can you tell us about what you’re working on these days?

Right now I am analyzing expression array data with R and try to understand the results. I’m also working on an extension of PDBFlex, and creating new UI features for Cancer3D.

If you were now starting your PhD, what 3 tips would you give yourself?

Try to collaborate outside your group

Try to be more independent in the early stages

Attend basic lectures rather than read books (German PhDs do (did) not have required lectures in my time)

Do you have someone you would like to thank that contributed to your research?

“JS is valuable for the future, especially if the wet lab does not work out”

Definitely! Everyone who was involved and had a significant contribution to the development of all the software we wrote together.

Thomas, thank you for this delightful interview. We wish you the best of luck deciphering the mysteries of the nanomachines we all love and may the Worm be with you!

Follow Thomas:

LW Thomas Harbe