Hi biostar people,
Kind of very busy these weeks. And I was wondering if there is some sort of concurrent versioning system for genome annotation. Most of my pals work from different timezones and places. Of course, it's always possible to use git or similar. But, explaning how such tool works (and why) to biologists can be rather cumbersome.
You'd think, wouldn't you, that version control and genome annotation go hand-in-hand. Yet a web search for those terms throws up very little.
One approach might be to use a wiki, e.g. the popular MediaWiki. Wikis feature version control, they often have plugins that allow extensive customisation and they have been used in other biological curation projects.
Some genome projects employ other tools for community annotation. A good example is the Paramecium project, which uses tools from GMOD. I'm unsure to what extent version control is employed.
The main web-based annotation software package that I know of is Manatee. It can store "old" annotations alongside "new", but I don't think it does true version control. It also requires quite a bit of administration to install (apache, mysql, multiple perl modules).
If you do want to go with just basic RCS, I think git could work. Subversion has a central repository, but anyone with rights could upload to it without notification. Using git, people could work in their own way then submit a pull request, which other members would approve before incorporation.
I don't think there is simpler version control than what is already available.
I found that svn was easy to explain to nontechnical users, especially using it via TortoiseSVN. I would avoid distributed systems such as git or mercurial.
It don't need to be simple. Biologists usually think of curation when doing version control. Curation can be rather cumbersome as well. In my imagination there is a system that is concurrent/distributed able to deal with tickets/milestones and community approval. Maybe just creating a "relational" layer on top of git to translate curation/annotation into versioning . . .
centralized systems are easier to understand. And they would need a centralized main repository anyhow. I have learned to never underestimate complexity, one tool may be substantially more powerful than the other, but a tiny bit of extra complexity may make it far less usable - thus far less effective in the big picture.
Questions like this often end up being answered with something like "You can use tool X". While that may be true, it is very important to define the use cases, work flows, and other requirements (web-based versus command-line, for example) before it is possible to evaluate answers. Spend some significant time planning and discussing some of these points with users/participants. Then, it will be clearer where to go next.
In my mind I always imagine that people will come up with an application X and an example on how to use it. Of course, I consider that most participants here are computer science guys or similar. What I really want to know is if there is a git-like system for genomes and where I can find it.
I've seen git used.
http://github.com/michaelbarton/Pseudomonas-fluorescens-R124-genome
While in principle a system similar to that used for Open Blogging should work, I am not sure that biologists can do this unless presented with simple templates on top of the platform. Personally, I'd use a wiki. Easier to use, understand and adopt (and that's hard enough). E. Coli Hub is a good example
Semantic MediaWiki is pretty straightforward to setup and seems like it has a lot of the features you're looking for.
I'd like to see an annotation interface like Google Wave with an integrated Google Maps style (i.e. non-Flash JavaScript) genome browser. Google Wave does versioning using XML-based operational tranformations.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
would something like Dropbox work? https://www.dropbox.com
Not really. It need to be concurrent for real, in the spirit of git. In such a way that we could test and commit versions of the annotated genome made with rather different pipelines.
I vaguely remember attending a seminar from an OBO/ChEBI curator who was explaining how do they were doing the annotation versioning. However, I can't find any link to this now... I am sorry, it was really a long time ago :-)