Juliette Wipf
Picture: Jason Banfelder, Director of the RU High Performance Computing Systems, talking about the most commonly used computing tools at the inaugural meeting of the SciComp group.
On April 12, Scientific Computing Users Group (SciComp) of The Rockefeller University’s (RU) held its inaugural meeting in CRC 406. The founders of the group, Jason Banfelder, Director of the RU High Performance Computing Systems (HPC), and first year graduate student Jazz Weisman, led the meeting. I caught up with Jazz Weisman about this new group on our campus.
NS: How did you and Jason come up with the idea to start the SciComp group?
I attended Jason’s Quantitative Understanding in Biology course at Cornell University and wanted to learn more. When I asked him about opportunities he said that starting a group is always a good, as well as a feasible idea. In fact, he had thought about starting something for a while as well. I actually recommend Jason’s lecture, or a similar intro level data analysis class, to everybody. A lot is already going on in that area, and we tried to create something in this pool. The future is definitely more computed, and we have to start somewhere.
NS: What do you think is the biggest plus of the SciComp group?
Painful and repetitive work should be reduced as much as possible. So many things can be done a lot easier with the help of computing, which will make repetitive tasks in science a lot less painful. But there are a lot of side benefits to our group. People get to know Jason as a representative of the IT department, which will make communication between the scientists in the lab and IT easier. People tend to be a bit shy about their computer skills, and we hope to make the IT department more accessible. Finally, we want to get interested people together. Labs can sometimes be a bit insulated; however, their computational interests would be similar.
NS: Researchers (myself included) can sometimes be a bit scared of using new programs, even though we use computer programs daily. Why do you think that is?
I think most are afraid of messing up their data. We also don’t want the design of our results to change, since we have long chains of experiments, sometimes generated over years, and a change in the output can sometimes make it hard to represent data neatly. But, as I said, most of our experiments come in long chains. Programming languages, such as R, Python or MATLAB, can simplify such tasks, and are actually a lot faster and easier to use than, for example, Microsoft Excel. Most importantly however, they make things repeatable, which is always better. If we use code to perform a string of tasks, this code can be given to a new student for example, and everybody can be sure the desired analysis was executed exactly the same way as usual. The student, on the other hand, can also study the string of code in peace and quiet, which will make understanding of the method easier for the new student as well.
NS: What can people expect from those meetings? Are there exercises that you do on computers together, or is it more of a discussion round?
Our group meetings usually start with a short talk of approximately 15-25 minutes on a chosen topic. For example, in our second meeting on May 18, we chose to talk about the data visualization tool ggplot2. After the presentation, we hope to get an open discussion going where everybody can ask questions. You can bring your laptop because it can help showing others the actual problem you are experiencing. It is not necessary that you attend the whole meeting; you can also just come for one part of it. We want our meeting to be an open thing. Also, we understand that everybody is busy and that you might have limited time for stuff.
NS: Who can attend the SciComp meetings? What skill level is expected from participants?
Absolutely everybody can attend our meetings and no previous experience is required. If you want to learn more on the discussed topic, please come. We expect nothing and are simply happy you are interested. If we talk about an R-based tool like ggplot2, for example, it will all make a bit more sense to you if you know some of the programming language R already. But it is not expected at all. We want the group to be widely accessible. Everybody who wants to should come!
NS: What do you expect from the participants (ask questions, prepare, etc.)?
People shouldn’t be afraid to get a discussion going. We are happy to answer the most basic questions! This is exactly why we thought the group environment would be nice, just to make everything more laid-back and relaxed. Ultimately we hope to also see group members helping each other out, with me or Jason only assisting when needed.
NS: What topics will be discussed in the meetings?
People can actually vote on which topic will be discussed. In this Google group, people should add their requested topics. If you and your colleagues want to learn about a specific program your lab is using, you should individually log onto the Google group and vote, so we can see how big the demand is. With this approach, reruns of hot topics are also possible if needed; just reenter the topic into the Google group. We hope to soon talk about DNA or RNA sequencing, which I definitely think is the topic most people are interested in at the moment. In addition, we will use the Google group for general updates as well as a place for people to ask questions.
NS: In your inaugural meeting, you talked about the most successful tools currently available to get a feel for the needs and interests of the attendees. In the last meeting you discussed the R plotting tool ggplot2, which makes all kinds of beautiful plots and graphs. When will the next SciComp meeting take place and what topic will be discussed?
We’ve decided to have the next meeting on August 3 in CRC 506 from 5:30 – 6:30. We will discuss Dynamic documents in R, presented by Thomas Carroll, head of the new bioinformatics resource center. Finally, if anyone is interested in becoming a co-organizer they should contact me via email at jweisman@rockefeller.edu. I think that one or two more people to plan and put the word out could be a good thing for the SciComp group.
