| Having working in the bioinformatics industry as an SE for 9 years I can both agree and disagree. 1. I agree that SE standards and good coding practice are completely absent in the bioinformatics world. I remember being asked to improved the speed of some sequence alignment tools and realized that the source code was originally Delphi that had been run through a C++ converter. No comments, single monolithic file. The vast majority of the bioinformatics code I worked with was poorly written/documented Perl.
In addition a lot of bioinformatics guys don't understand SE process and so rather than having a coordinated engineering effort, you end up with a lot of "coyboy coding" with guys writing the same thing over and over. 2. I agree that productivity is very slow. This is a side product of research itself though. In the "real world" (quoted) where people need to sell software, time is the enemy. It's important to work together quickly to get a good product to market. In the research world, you get a 2/5 year grants and no one seems have much of a fire under them to get anything done (Hey we're good for 5 years!). You would think that the people would be motivated to cure caner quickly (etc), but it's not really the case. Research moves at a snail's pace - and consequently the productivity expectations of the bioinformatics group. 3. I disagree that research results from the scientists are garbage. Yes it's true that some experiments get screwed up. However, if you having a lot of people running those experiments over and over, the bad experiments clearly become outliers. Replication in the scientific community is good because it protects against bad data this way. Somehow the author must have had a particularly bad experience. 4. Something the author didn't mention that I think is important to understand: most scientists have no idea how to utilize software engineering resources. The pure biologists, many times are the boss, and don't really understand how to run a software division like bioinformatics. Many times PHD's in CS run a bioinformatics group, who have never worked in industry and don't know anything about good SE practice or how to run a software project. A lot of the problems in the bioinformatics industry is directly related to poor management. Wherever you go you're going to have team members that have trouble programming, trouble with their work ethic, trouble with following direction. However, in a bioinformatics environment where these individuals are given free reign and are not working as a cohesive unit, you can see why there is so much terrible code and duplication. |