In the correspondence section of Nature this week is a suggestion for a gene function wiki. Not a bad idea. We definitely need a single, integrated, comprehensive and authoritative resource of this kind for genome data. There are standalone alternatives, but why have individual groups compiling their own annotations when we could be sharing?
Unrelated - I don't think we announced officially that Nodalpoint now features forums. Follow the link at top right.
Comments
More than just wiki is needed...
Hi All,
As a way of introduction (I am a newbie within the nodalpoint)...
My name is Marc A. Marti-Renom, I work at UCSF and am very involved with two BioMed open-source projects:
In particular, I run the Malaria Research community of the TSL (see more here). We have been thinking about several projects we would like to carry at TSL/Malaria including a sort of wiki page for genes in the malaria genome. We think that now is the time to start getting this things coded and getting researchers to "play" with the community based tools that are very popular in other fields (ie, blog, wiki, tagging, etc...).
We have envisioned what we call the "GeneCards" that will collect all available information (mostly from NCBI) about a given gene. Then users of the community would be able to do several things with the cards:
i) VISUALIZE. Users will be able to see what a gene card has to say about a given gene. Imagine being able to visualize centralized data for a given gene.
ii) MODIFY. Users will be able to add information. Imagine a GeneWiki specialized for biological data entry.
iii) LINK. Users will be able to (using bookmarklets) link pieces of information in the web to a given gene (including pubmed entries). Imagine Connotea/CiteUlike with a third dimension (user/data/gene).
iv) "OWN". Users will be able to save a GeneCard into their "baskets" and associated data to them (such as tags). Imagine Flickr for genes.
Of course, all of this is just in our imagination for now (and still for some time). We are planning to develop de needed modules within Drupal and initially apply them to the malaria genome. Any help is welcome! ;-)
I think that the opened discussion in the NodalPoint is a good starting place for shaping the most useful of such a web-app for BioMed research. I hope that we can all find a good solution to the more general call of getting biomedical researchers into the 21st century web tools
Best to all!
marc
GeneWiki started by Tom Stambaugh
Pierre posted in his blog a link to a new wiki page.
not just 'a wiki'
There's a big difference between 'a wiki' and 'a atructured database that anyone can edit'. Unfortunately this wiki page is just 'a wiki page' - there's nothing that makes it particularly useful for gene function annotation.
Something like manatee
True enough. The closest thing that I can think of is Manatee from TIGR. It's quite large and complex, designed for collaborative manual annotation of genomes by group members.
Something similar could work for a gene "wiki". A structured database (perhaps loading GenBank or GFF files via Bioperl tools or similar), a web frontend where users register and login, then structured forms that map to the DB schema. I thought about doing this in our group but had no time, plus we have manatee experts on hand.
The nature letter was also
The nature letter was also posted on the Semantic Web for Health Care and Life Sciences mailing list (what a mouth full). The discussion is on going, with some interesting points being made by Eric Jain (Uniprot in RDF) and Matthew Cockerill (Biomedcentral). As I suspected our bio-ontology-overlords are skeptical. This misses the point that symbols only have meaning in the context of a community (i.e. shared meaning). Anyone building an ontology is just codifying meaning as accepted by the community at one instance in time.
I also discovered that Eric Jain has a weblog with an interesting piece on how to encorage contributions (in this case to Uniprot). His blog will be duly added to the ever growing list of recommended reading (same goes for Pierre's blog Yakafokon).
amazon.com for genes
>why have individual groups compiling their own annotations when we could be sharing?
We need to do both. I wish gene databases like SGD (yeast) offered users the ability to comment on genes, the way amazon.com allows users to comment on books, and make lists of books. (BTW, I wrote a commentable and concise yeast gene database 7 years ago, that went almost completely unused - so I know how these projects can fail without momentum). People need a place to store gene lists, and curate their own annotation about genes, and choose whether those annotations are made public or not. One benefit would be that we could use these annotations to computationally iterate through data, similar to doing gene set enrichment analysis (e.g. Broad et al.), to then discover which annotations are informative, and which are not.
CAPs
An alternative is community annotation as was done for Pseudomonas aeruginosa. Divide the work between expert researchers and rank annotations as certain (experimental evidence), very likely (strong computational evidence), possible (weak computational evidence) and unknown (no evidence at all). I thought this was a great approach but it was not widely adopted. Microbial genomes started pouring out at such a rate that people turned to comprehensive automated annotation pipelines instead. The two can coexist - do a good quick auto, then a slower manual with frequent announcements as to how much has been manually examined. Sounds like a wiki to me.
GeneWiki
There is already a GeneWiki. However, it seems not to be active, so no one is using it. It lists genes (e.g. CCR5) as well as diseases (e.g. Agammaglobulinemia).
I think I came across
I think I came across GeneWiki before, but thanks for pointing it out. I am all for wikis as a way of doing collaborative science, my initial thought was to use wikis to build ontologies, however that would require some kind of formal link structure between pages. Regardless I think that the reasons for wikis not being actively used are: trust and awareness. The very thought that some random scientist may have edited information about a gene may scare some users. It would have to be peer reviewed by experts before people trust it right ? That is why peer review works (sarcasm intented).
I find it interesting that this idea was the subject of a letter to Nature and a GeneWiki already exits. Does this further fuel the idea that most biological scientists are out of touch with the web ?
Spitshine (Notes from the Biomass) has pointed out that the medical scientists are a head when it comes to utilising online tools for science communication.
Not in this case
Does this further fuel the idea that most biological scientists are out of touch with the web ?
For once, I think not :)
If they know about wikis, they presumably did their research before writing the letter. WikiGene front page hasn't been updated in over a year - they probably discounted it. Plus from its description, it's not exactly the idea they proposed (more to do with regulatory networks).
Perhaps WikiGene would get more attention with a better domain name - andromeda.gsf.de?
Wikis are sometimes too complicated
I think the lack of confidence is one reason. Another reason might be that the syntax of Wikis might be too complicated for "normal" users, especially if the content is some kind of structured content as in the GeneWiki.
Another difficulty, which, however, affects only the server-side: How to synchronize the content of the Wiki with e.g. databases like Swissprot, etc. For example, much of the content of the GeneWiki is generated content, so it must be updated regularly. However, the user changes must be incorporated, too. This only works fine if the user does not violate the special formatting/annotation rules and so the user has to learn these rules first. Not a really good solution.
What is needed is something like Structured Blogging for Wikis. So you can define what the structure of a page should look like and the user sees only the content and also can only edit the content but not the structure.
MS has done something similar with InfoPath. You have an XML Schema that describes your data. From this schema you generate your input form and using this input form you can generate XML files.
I've already searched for a XML-Wiki or something like that, but didn't find such a software. I know that TWiki supports forms.
Wiki for genetic disease
I agree. A wiki for information on genetic disease would be helpful, too. Then problems like the one that I wrote about here, where someone mixed up alanine and arginine, could be fixed in a timely manner.
Sandra Porter
Discovering biology in a digital world
Here's another application of a wiki for genes etc.
3Wiki
a wiki for the type three secretion system used by pathogenic bacteria to translocate proteins into host cells.
Setting up a Wiki for gene names is so obvious, I wonder how it made it into Nature. I am sure they have played with Wikis themselves.
That would be a nice idea
That would be a nice idea for writing comments and annotations about genes, proteins, etc... but that would be a problem for knowledge discovery. We already have this problem for pubmed: there is already a large number of tool used to parse and analyse the abstracts in the papers (see: http://dx.doi.org/10.1038/nrg1768 for a recent review) in order to generate biological pathways, etc.... The future of annotation is in RDF/semantic web.
Pierre
I believe you meant to link
I believe you meant to link to http://www.nature.com/nature/journal/v439/n7076/full/439534a.html :-)