Annotation Conf. Call, March 11, 2014
SGD: Rama, Selina, Stacia, Diane, Edith
MGI: Karen, Li
UCL: Ruth, Paul, Ana
EBI: Rachael, Jane, Aleks, SusanTony
Berkeley: Chris, Moni, Suzi
Hardware changes (Rama)
We moved many of the GO services (website, SVN, CVS etc) to a new virtual machine (@Stanford).
- GO SVN repository has been moved to this new machine. If you are having trouble checking your GAF or any other file into SVN, please send an email to GO-admin: email@example.com or to firstname.lastname@example.org.
- The GO wiki has also moved to the new machine, and the software has been upgraded. The GO wiki now requires that every page has at least one category. You will not be able to create a new page or successfully edit an existing page if it does not contain a category. Existing GO categories are listed at: http://wiki.geneontology.org/index.php/Special:Categories. More information about categories is available: http://www.mediawiki.org/wiki/Help:Categories.
From MGI (Karen):
I am annotating some genes to the component term (colocalizes with) "nuclear chromatin". I would like to indicate a specific kind of region within nuclear chromatin, specificlly RNA polymerase III promoters, with a SO term: "RNApol_III_promoter ; SO:0000171" I am already using the colocalizes qualifier since the experiment is a ChIP, but I don't see what existing relationship I could use in order to include a SO term to describe the specific types of regions within the chromatin.
Differential Expression Experiments and IEP Evidence Code
From Kimberly VA (WB)
- Background: We recently had a WB user request that we annotate some genes based on the results of differential expression experiments, specifically microarrays and qRT-PCR.
- In the past, we've been very conservative about annotating to Biological Process terms from differential expression experiments, but this request has raised the issue again and we are re-visiting our annotation practices.
- In this particular case, these are experiments testing the transcriptional response to infection with pathogenic bacteria although we have a number of similar papers and experiments that examine the transcriptional response to various other stressors, such as oxidative stress, thermal stress, etc.
- The authors have asked that we annotate the statistically significantly expressed genes to 'defense response' and 'innate immune response'. Note that we had already annotated genes from this paper for which the authors performed follow-up RNAi experiments to try to assess functional significance.
- Many of these genes do not have other GO annotations and the gene products do not have domains that yield annotations based on sequence aimilarity.
- The paper in question was published in 2006, but using Textpresso to search the literature, we find that a number of these genes are considered biomarkers for infection response, even in the absence of functional data.
- We are considering annotating to a high-level response term, e.g. 'response to bacterium', but not to the more descriptive, and functionally suggestive, terms of 'defense response to bacterium' and 'innate immune response'.
- The issue of annotating based on differential expression experiments has come up before in GO. What are groups currently doing and why, and what, if any, feedback have groups gotten from users wrt these types of annotations? Does GO need to provide more annotation guidelines to curators for this area of annotation?
- SVN: Jane reported some SVN issues. They were resolved by Stuart (Admin at Stanford), see details in Stuart's email below-
Hi Jane and Paola, Can you please give more details about the problems you are having with SVN? What command are you trying to run? Are you trying to commit changes or just check out files from the repository? One quick possibility that comes to mind – are you using an automated script to make an authenticated connection to the SVN server? If so, you'd have to modify your .ssh/known_hosts file in your home directory. You'd want to remove ext.geneontology.org from that file, and then try running a command like ssh email@example.com date which will create a new entry in your known_hosts file. Until you do this, SSH will complain that it can't verify the authenticity of the server and it won't allow the connection. And if you are connecting from an automated script, depending on how the script is written, you may never get the output saying that this is the problem. If this is not the problem, please feel free to send the command you are running and any output errors in an email to firstname.lastname@example.org and we'll debug the issue further. Thanks, – Stuart
- GOC wiki: If the page doesn't load, please clear cache and reload. Please don't add new categories. Pick up from one of the existing categories. Currently the Wiki is set up to issue an Warning, but soon Category will be required to save a page.
- None of the relationships we have currently will work for this case. We will put a term in the ontology instead (Karen will make a SF ticket). Chris will talk about it in the Ontology meeting on Thursday.
- Kimberly: Will you make a Chromatin binding annotation for this data? Karen: I don't make it for ChromatinIP. No binding annotations for ChIP.
- If something is upregulated it doesn't mean the gp is actually involved in the process. There can be false positives.
- Tanya: In the absence of other information, annotation to one of the 'response_to X' terms might be useful to the user
- Suzi: If we want to provide a lead or cue to the user then the annotations are addressing the needs for a different community. There are users who want rock solid experimental annotations and they will be disappointed with these 'response_to' annotations
- Ruth: there are lots of IEP studies in human and we could potentially be making lot of these annotations. But then they can get circular and be propagated. For this reason we don't want to be making these annotations.
- If you don't normally annotate data from Northern blots then this shouldn't be used either.
- We don't want to flood GO with these annotations.
- We should direct users to other tools to mine this data.
- you could use this data in conjunction with ISS or some other ev. code (from multiple papers) though. We cannot string evidence codes yet, but we can talk about it at the GOC meeting.