Add links to gene product list for sample frequency count in RTE (term enrichment)

[rbalakri]

Hi,

In TE results page the Count in the Sample frequency column is not linked to the gene names that contribute to that count. I think it is very useful to know which genes in your input list are enriched for that term.

Thanks,

Rama

[kltm]

AFAIK, the current TERF does not specify this, so it is not there. Personally, I'm a little worried about creep here.

While useful, I think we also need to be careful of the scope of the services we offer (and the scope of the protocol as it needs to be generalizable) and have a set way of forwarding people to the originating engines' pages (if they exist). If we define the protocol with an expanding feature set, we'll make it harder for more providers to adopt and we'll have less engines overall.

features >?< number of providers/speed of development

Input from @cmungall and the wider group would be useful to set the standard.

[rbalakri]

Seth,
AFAIK authors are concerned about which genes are enriched for a certain term, not the #. Personally I think it is a must have!

I am not qualified to comment on your other comments about other engines adopting etc.

We need input from the larger group not just on this feature request, but other missing features in this instance of TE. I am happy to bring them up at Barcelona or your software meeting.

Rama

[kltm]

Again, we'll need further input from @cmungall, PANTHER, etc. At the end, this is only a (relatively small) AmiGO issue at the end of a larger process:

• get input for the requirements of a minimal TE analysis that is most useful to the broadest range of our users
• modify TERP to reflect this
• implement this new TERP at PANTHER
• consume the new data in AmiGO (which is this issue)

The last step in there is not that large, as we'll just be doing a little additional processing to the return data (although keep in mind we have work to do with #111 and #117).

It would probably be useful to have our own reference implementation of TE for others to look at and copy when implementing TERP (but Solr 4.x, etc.).

This might be best covered at the meeting to get the most correct input, with the groundwork done beforehand with the major players.

[cmungall]

Note that TERP needn't be extended here (extending it is a reasonable idea but this should be optional as we don't want to send high volumes over the wire).

Everything is in amigo, clicking on the sample frequency column value should take you to a gene query that shows all genes in the GO term intersected with the sample.

The implementation could be fiddly. One way is to create a solr query

term_closure=GO:nnn AND (bioentity_id= OR (bioentity_id= ... )

But may not be possible in our solr config.

the other option is to do the intersection outside of solr. Use solr for one or both of the conjunctive clauses, then intersect in code. Have to watch out for limits.

[kltm]

I think that this kind of query would be fiddly and possibly not possible for larger sets of GPs, although we could try the experiment.

I think that this may bring us to the issue of what will happen when people (reasonably) want this data for download for the whole set. We've thus far avoided iterative queries (although #69 will change that), since they can cause performance issues and are not really in line with how the index should be used.

Really though, isn't this just essentially a slimmer tool for AmiGO? If we had this generally solved, we could plonk it in there.

[vanaukenk]

Hi -
Just adding to this thread since I've spent some time with a user over the past week or two who is trying to get a list of the gene products annotated to specific terms in their list of enriched terms.

If at all possible, I think it really would be very helpful to users to be able to click on the number in the sample frequency column and get the list of genes/gene products. Alternatively, or additionally, listing the genes after the number in the sample frequency column would be helpful, too.

Being able to see the background gene list would be helpful, as well.

--Kimberly

[rbalakri]

Hi Seth, Chris,

I think you told me that you have a fix for this issue already and is ready for testing. There are few GO help emails asking for this feature (i.e. getting access to the names of the genes for each enriched term). can we move this to production or testing servers?

Thanks,

Rama

[kltm]

Rama,

The current release is on hold while loading issues are being dealt with in production.
As for this specific issue, the "fix" in this case will be to bypass the AmiGO interface completely for the time being and just forward users into the pantherdb.org site.

Cheers,

-Seth

[kltm]

For the time being, we're adding a bypass through the amigo RTE and directly into PANTHER (amigo still being used here to fix the input before sending the user to PANTHER). The earliest we'd get back to this is 2.3.