21 Oct 2008--bio-to-CS mapping
Mark Braunstein introduced himself at the start of the meeting. He's here in the College of Computing as a Professor of the Practice to help create a joint School of Biomedical Informatics, between GT's College of Computing and Emory's School of Medicine.
First item to decide in mapping the Biology projects from the last month to CS activities is the language. Any given language has affordances and challenges which will determine the complexity of a given domain project when mapped into that language.
- We decided not to pick MATLAB. Too few biology groups use MATLAB, so there is a lack of support, and the skill isn't directly useful in biology projects. MATLAB is mostly used for population modeling. There is more use of R in Biology.
- The key factors for the Biology faculty were (1) readability and (2) the ability to go on, to build on the given learning and to use the skill in the given language in other projects.
- Monica and Mark talked about how the Python->Java transition works. CS1315=>CS1316 (both media computation) really works, probably because the domain remains the same when the language changes. CS1301=>CS1331 works, but the challenge is picking up objects in Java. It's not a "syntax trainwreck" though. In general, the Scheme=>Java transition did not work, for reasons we don't really understand.
- Whatever we decide for BioCS1, what comes next is probably Java.
- Perl is the legacy tool of choice in biology. However, it's not readable.
We finally settled on BioPython.
Attached is the document describing our first pass mappings of Biology projects to CS implementations, and how these would play out in an intro course.