Slashdot Mirror
Open Source And Genetics
UnanimousCoward writes "SFGate has an article about some researchers pushing for the open sourcing of genetic research software. Of course, the pros and cons are debated." It's the age-old debate; what follows the heart of the scientific method more? Peer review, or getting the information out as fast as possible?
Didn't we see this article yesterday right here?
This is just my opinion (YMMV, IANAS, etc), but I would argue for peer review, insofar as it relates to the ability to duplicate results and experiments. My thinking here is that the ability to duplicate results is at the heart of science. If others can't duplicate them, the results aren't accepted.
A grain of salt is provided for those who disagree.
The Independent: Reverend Spooner Arrested in Friar Tuck Incident - ISIHAC, Historical Headlines
I don't think genetic code and source code should be treated the same. After all malicous sourcecode only hurts computers and inconvienices users. Genetic code effects us on a much more real level.
I can release code that has a bug in it and fix it pretty quicky. If it causes problems, it only causes problems. Meddling with life is not so trivial and the concequences can be fatal. It is for this reason that I am against open source genetics.
Slashdot's rate-of-post filter: Preventing you from posting too many great ideas at once.
If public funds were used to finance the research, then it should be public information. The only exceptions should be driven by security concerns. If private monies are put into the mix, let us thank the donor for their donation.
We are the public and our tax dollars are used to generate this information. We own it.
Give it to us at the bare cost of distribution.
What should be open source and free beer free is the specialized software used by many researchers for processing DNA and Protein sequence information. For instance, take the very nice software package Vector NTI. Does many things researchers want. Also costs $5000. Instead of NIH (government science funding agency) money being spent many times (once per funded lab), what they should do is earmark some money to get programmers to write for free usage software like Vector NTI. It would cut down on the redundant purchase of software by many government-funded laboratories. Don't even get me started on how much research money is being spent on Windows, Word, Excel, PowerPoint, and Adobe Acrobat...
Good research is best accomplished by having all access to all information, so that you don't have to re-invent the wheel or relearn the experiments.
So why do some companies reproduce research that 10 other groups/companies are doing?
1. To see if the work can actually be reproduced. Patents are notoriously unreliable for their reproduciblity in the chemical sector.
2. To learn more about the technology, to either improve it or use it for another project. By doing the experiments yourself, you get a lot of knowledge that is never written down on paper.
3. Having a completely different group of scientists working on a research project can lead to a completely different way of interpreting the results. Therefore, the conclusions and applications of the research are different. (Diversity of thought!)
4. The idiots forgot to do a literature search before starting their project. This unfortunately happens more often than it should.
It does look like at first glance that reproducing other previously done work is a waste of time. However, given the advantages of the first 3 points, it will continue to occur this way, and for the benefit of science, it probably should.
-When going for broke, go for Ithaca!
Geeting info ASAP for review is the only REAL peer review. What journals call ``peer review'' is just editorial censorship of any real advance in science; any revolutionary science is deemed ``not suitable for publication''. Or sent to refeeres chosen by the *editor*. Any paper that does not conform to the prevailing views is NEVER published in major journal---want to check? Only more-of-the-same.
So, as a scientist myself, I vote for fast release: let ME (a real peer) do the review. Editorial censorship? Thank you very much, but no, I stopped needing paternalism after my early teens.
Gustavo A. Concheiro Perez
There was an interesting discussion on this topic last year over on bionet.software, with contributions from several authors of significant bioinformatics software (search for 'open source' in the bionet.software group at groups.google.com).There's also a good interview with Ewan Birney (of BioPerl and EnsEMBL fame) about open source bioinformatics here:
http://conferences.oreilly.com/biocon/
Like some of the other contributors to this thread, the argument I find most convincing is that peer-reviewed scientific publications usually (and for very good reasons) require full disclosure of methods - why should software be exempt from this? Similarly valid criticisms were made about _Science_'s decision to publish the Celera human genome paper with only conditional access to the primary data. Biotech companies are of course free to conceal both their data and methodology, but the scientists involved should not then expect the right to contribute to the peer-reviewed literature.
There are also sound economic reasons for an open source approach to significant scientific software: it's ironic that software developed with public funding has in some cases been commercialized by the host institution of the programmers, leading to a situation where other laboratories that receive grants from the same funding body end up paying substantial license fees! (some funding agreements are now savvy enough to include a clause preventing this). In fields like molecular biology, there's also the risk that initially reasonable license fees can rocket when the software proves to be useful. This is pretty much what happened in the case of GCG, a set of sequence manipulation utilities originally developed by a university department, but later acquired by a large biotech company. The increasingly restrictive and expensive license was one of the factors that led to the creation of EMBOSS:
http://www.hgmp.mrc.ac.uk/Software/EMBOSS/
a GPL/LGPL'd alternative that in many respects now surpasses GCG, and runs on a wider range of Unix-like platforms (including Linux, Cygwin and MacOS X).