Re: [intact-developers] Unable to query genes anymore using REST interface

11 views
Skip to first unread message

mar...@ebi.ac.uk

unread,
Feb 24, 2013, 8:06:19 AM2/24/13
to intact-d...@googlegroups.com
Hi,

We updated our service to export in MITAB 2.7 so we changed some fields :

- proteins will have uniprot ids and intact acs in unique
identifier/alternative identifiers columns
- genes will have ensembl and intact acs in unique identifier/alternative
identifiers columns
- nucleic acids will have DDBJ/EMBL/genbank and intact acs in unique
identifier/alternative identifiers columns.
- small molecules will have CHEBI ids and intact acs in unique
identifier/alternative identifiers columns.

All the other xrefs (such as flybase, interpro, go, ..) will go in the
xrefA and xrefB columns of MITAB 2.7 (columns 24 and 25).

If you want to be sure to collect identifiers and xrefs (for proteins,
ENSEMBL ids will be in xrefs), you can use this query instead :

http://www.ebi.ac.uk/Tools/webservices/psicquic/intact/webservices/current/search/query/identifier:FBgn003922
OR pxref:FBgn003922

However, we don't have any results for this flybase identifier so far.

Don't hesitate to contact us again if you have more questions about our
PSICQUIC service.

Regards,

Marine

> Hi,
>
> I have some scripts to retrieve interactions starting from ENSEMBL IDs
> using the rest interface. I've tried them today after a long time and
> notice no interactions are retrieved anymore. The return code is 200 so
> the
> problem is I think with my rest string.
>
> This is an example of what I use
>
> http://www.ebi.ac.uk/Tools/webservices/psicquic/intact/webservices/current/search/interactor/FBgn0039228
> AND type:"physical*" AND detmethod:"experimental*"
>
> Has something changed wrt the string syntax? Thanks a lot for your help.
>
> Best
> T
>
> --
>
> ---
> You received this message because you are subscribed to the Google Groups
> "intact-developers" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to intact-develop...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>


TRypdal

unread,
Feb 25, 2013, 5:31:55 AM2/25/13
to intact-d...@googlegroups.com
Hi Marine 

thank you for your reply. Yes I have more questions.

1 What is now the correct usage for wildcards in the query? For example, I might be interested in getting all direct binary experimental interactions and rejecting interactions from spoke expanded complexes. This won't work


while this does


But I'm not sure whether these are considered semantically equivalent

2 How likely is it that the standard will change again in the near future? How likely is it that future changes will impact the query syntax language? I ask because these might impact anyone who is planning on using the interface for significant work (scripts/modules/programs). 


Thanks a lot for your help

Marine

unread,
Feb 25, 2013, 6:58:06 AM2/25/13
to intact-d...@googlegroups.com, TRypdal
Hi,
Hi Marine 

thank you for your reply. Yes I have more questions.

1 What is now the correct usage for wildcards in the query? For example, I might be interested in getting all direct binary experimental interactions and rejecting interactions from spoke expanded complexes. This won't work


while this does


But I'm not sure whether these are considered semantically equivalent
When you use a wild card query, you should not put it between quotes. You can use this query instead :
Your second query is very similar to the first one as we will return any results having the 'physical' word in the interaction type and the 'experimental' word in the detection method. If you use physical*, it will not only match all the results having the 'physical' word in the interaction type but also all the results having a word starting with physical. The best example will be with aliases where brca2 will match any aliases having brca2 but brca2* will also match any aliases starting with brca2 (such as brca2b).

For querying all binary interaction and rejecting spoke expanded, you can now use the complex field in MIQL 2.7 : http://code.google.com/p/psicquic/wiki/MiqlReference27. For instance this query will return all binary interactions that are not spoke expanded and have identifier FBgn0010433 or xref FBgn0010433
http://www.ebi.ac.uk/Tools/webservices/psicquic/intact/webservices/current/search/query/identifier:FBgn0010433 OR pxref:FBgn0010433 AND complex:"-"



2 How likely is it that the standard will change again in the near future? How likely is it that future changes will impact the query syntax language? I ask because these might impact anyone who is planning on using the interface for significant work (scripts/modules/programs).
It is unlikely that the MIQL query language syntax will change again and not be backward compatible. We may want to keep extending the MIQL query language as we are doing now with MIQL 2.7 but it will only add new fields, not remove fields. The current fields to query by default in a free text search in psicquic are identifiers and aliases, pubId, pubAuthor, interaction_id, detmethod, type. We may want to extend the default fields for MIQL 2.7 (adding pxref and xref as default fields for free text search so it will pick up FBgn0010433 without having to specify the query field but that needs to be discussed more in details in the PSI meeting with all PSICQUIC partners) but we don't plan to remove any default fields for backward compatibility.

We are currently working on defining and implementing the Data best practices in PSICQUIC (see http://code.google.com/p/psicquic/wiki/DataDistributionBestPractices), which may require some changes in where we put the information in the MITAB columns so all PSICQUIC services look the same. In IntAct, that explains the main changes why the xrefs have been moved from identifiers to xrefA and xrefB columns for instance. But the data best practices are now better defined and it is unlikely that you will find big changes in IntAct PSICQUIC webservice such as xrefs moved to another column (so they are not indexed as identifier but as pxref).

Regards,

Marine


Thanks a lot for your help

On Thursday, 21 February 2013 11:48:35 UTC, TRypdal wrote:
Hi,

I have some scripts to retrieve interactions starting from ENSEMBL IDs using the rest interface. I've tried them today after a long time and notice no interactions are retrieved anymore. The return code is 200 so the problem is I think with my rest string.

This is an example of what I use


Has something changed wrt the string syntax? Thanks a lot for your help.

Best
T
--
marine.vcf

TRypdal

unread,
May 7, 2013, 1:11:16 PM5/7/13
to intact-d...@googlegroups.com, TRypdal
Hi Marine

thank you for your explanation. What is the correct wildcard to retrieve all interactions tagged with "physical association" or any of its specialisations? I used to be able to use "physical*" and this would retrieve interactions that at any level of the PSI-MI obo ontology under physical association: 

physical association
direct interaction
....

same for "detmethod:experimental". Can you confirm that, like before, typing "detmethod:experimental*" will retrieve all experimental detection methods?

Thanks again

Best,
G

Marine

unread,
May 8, 2013, 6:33:27 AM5/8/13
to intact-d...@googlegroups.com, TRypdal
Hi,

If you want all interactions having interaction type = "physical association" or any of its children in the PSI-MI ontology, you can simply ask :

type:"physical"

This will retrieve interactions having type = "physical association", type = "direct interaction" and all its children.

This can only work in IntAct PSICQUIC webservice because we enrich the ontology and cannot work with other PSICQUIC webservices where the webservice is not aware of the ontology.

If you want all the interactions having detection method which is a child of experimental interaction detection, your query can be :

detmethod:"experimental"

This will return any interactions having 'experimental' word in the detection method and all the children. Again, this is only possible in IntAct because our service is enriched with the ontologies.

If you use type:physical*, it will also return interactions having type = "physical association", type = "direct interaction" and all its children. But it could also return any other type containing a word starting with physical (physical can only match the word physical and physical* can match physical, physically, etc.)

Cheers,

Marine
marine.vcf
Reply all
Reply to author
Forward
0 new messages