Hey Ed,
Thanks a lot. How much trouble would it be to cut a release for this?
Or, if you send me a diff file I can compile a new version myself.
If we use the current Apollo, the next step in our annotation process will
convert the ranks to 0, which I'd like to avoid - someone may have
annotated some trans-spliced genes (or may want to in the future).
-j
On Tue, Jun 10, 2008 at 4:40 AM, Ed Lee <[EMAIL PROTECTED]> wrote:
> Hey Justin,
>
> I've updated the Chado XML writer to handle exon ranks in a more correct
> manner. Also fixed a "bug" where ranks generated were 1-based rather
> than 0-based (look at the original Chado XML file you loaded into Chado
> - the ranks are 1-based rather than 0-based). Please keep in mind that
> the adapter was written specifically for FlyBase and is by no means a
> general purpose Chado XML adapter (which we should probably write at
> some point down the line). I'd imagine that this is not a show stopper
> on your side, so you won't see the fix until the next release. But if
> this is urgent, let me know and I'll cut a release.
>
> To the Flybase folks:
>
> Hopefully these changes won't break anything on your side. I've noticed
> that the ranks in Chado XML were 1-based, which goes against the Chado
> convention. Was the rank of "0" used as some kind of place holder for
> your programs to replace? Let me know if the changes will break
> anything on your side.
>
> -Ed
>
>
> On Tue, 2008-06-10 at 01:27 +0700, Justin Reese wrote:
>> Hi Mark,
>> Yeah, I don't anticipate needing the feature relationship rank
>> attribute for anything, I thought I should pursue it though because
>> 1) Some other piece of software down the pike might care about
>> rank
>> 2) it's kind of curious that Apollo would write the rank out
>> to Chado XML correctly but not read it from Chado correctly
>> correctly
>> 3) might help improve Apollo, if this turns out to be a bug
>>
>> -j
>>
>> On Tue, Jun 10, 2008 at 1:16 AM, mark gibson <[EMAIL PROTECTED]> wrote:
>> > This doesnt surprise me - as you see i think the chado xml adapter doesnt
>> > deal with rank. rank is something that apollo doesnt need and so its not
>> > tracking. in fact i dont think theres anything in the apollo datamodel for
>> > rank - and it would need to be added (either explicitly or through a
>> > property).
>> > Nomi wrote the chado xml adapter and she can probably elaborate further.
>> > The question though is why do exons need to be ranked in chado? Isnt it
>> > redundant with coordinates? Why does exon rank mean anyways?
>> > cheers - Mark
>> >
>> > On Jun 9, 2008, at 1:34 PM, Ed Lee wrote:
>> >
>> >> Hi Justin,
>> >>
>> >> Yes, please provide me with some sample data and access to the database
>> >> and I'll explore this issue further.
>> >>
>> >> -Ed
>> >>
>> >> On Tue, 2008-06-10 at 00:20 +0700, Justin Reese wrote:
>> >>>
>> >>> Hey Apollo-ers,
>> >>> We are using Apollo, Chado and XORT to handle the bovine genome
>> >>> annotation effort, and things are working pretty well.
>> >>>
>> >>> I am running some tests to make sure everything is working correctly,
>> >>> and it seems Apollo is not pulling some rank attributes out of Chado
>> >>> correctly.
>> >>>
>> >>> When I do a roundtrip test thusly:
>> >>>
>> >>> 1) Annotate gene using Apollo, save as Chado XML
>> >>> 2) Load into Chado using XORT
>> >>> 3) Pull annotation from Chado using Apollo
>> >>> 4) Write to Chado XML ("after" XML file)
>> >>>
>> >>> and do a diff on the "before" Chado XML (step 1) and "after" Chado XML
>> >>> (step 4), the rank attribute of feature_relationship's have changed.
>> >>>
>> >>> In the "before" Chado XML, the rank attribute of feature_relationship
>> >>> between
>> >>> exons and their mRNA in the before Chado XML is always equal to the order
>> >>> of the exon: 1st exon has a rank of 1, second exon has a rank of 2 and so
>> >>> on.
>> >>> In the "after" Chado XML, the rank attribute is always 0.
>> >>>
>> >>> Example:
>> >>>
>> >>> Before -
>> >>> <feature_relationship>
>> >>> <rank>1</rank>
>> >>> <subject_id>
>> >>> <feature>
>> >>> <is_analysis>0</is_analysis>
>> >>> <name>Btgn:temp1:Chr6.29:650000-768000:1</name>
>> >>>
>> >>> After -
>> >>> <feature_relationship>
>> >>> <rank>0</rank>
>> >>> <subject_id>
>> >>> <feature>
>> >>> <is_analysis>0</is_analysis>
>> >>> <name>Btgn:temp1:Chr6.29:650000-768000-RA exon 1</name>
>> >>>
>> >>> XORT seems to have correctly loaded the ranks into Chado, so that's not
>> >>> the
>> >>> problem:
>> >>>
>> >>> SELECT feature.name, feature_relationship.rank FROM feature,
>> >>> feature_relationship WHERE feature.feature_id =
>> >>> feature_relationship.subject_id AND uniquename LIKE
>> >>> 'Btgn:temp1:Chr6.29:650000-768000%'
>> >>>
>> >>> name | rank
>> >>> -------------------------------------+------
>> >>> Btgn:temp1:Chr6.29:650000-768000-PA | 0
>> >>> Btgn:temp1:Chr6.29:650000-768000:16 | 16
>> >>> Btgn:temp1:Chr6.29:650000-768000:15 | 15
>> >>> Btgn:temp1:Chr6.29:650000-768000:14 | 14
>> >>> Btgn:temp1:Chr6.29:650000-768000:13 | 13
>> >>> Btgn:temp1:Chr6.29:650000-768000:12 | 12
>> >>> Btgn:temp1:Chr6.29:650000-768000:11 | 11
>> >>> Btgn:temp1:Chr6.29:650000-768000:10 | 10
>> >>> Btgn:temp1:Chr6.29:650000-768000:9 | 9
>> >>> Btgn:temp1:Chr6.29:650000-768000:8 | 8
>> >>> Btgn:temp1:Chr6.29:650000-768000:7 | 7
>> >>> Btgn:temp1:Chr6.29:650000-768000:6 | 6
>> >>> Btgn:temp1:Chr6.29:650000-768000:5 | 5
>> >>> Btgn:temp1:Chr6.29:650000-768000:4 | 4
>> >>> Btgn:temp1:Chr6.29:650000-768000:3 | 3
>> >>> Btgn:temp1:Chr6.29:650000-768000:2 | 2
>> >>> Btgn:temp1:Chr6.29:650000-768000:1 | 1
>> >>> Btgn:temp1:Chr6.29:650000-768000-RA | 0
>> >>>
>> >>> I did some Apollo code diving, and found this in ChadoXmlAdapter.java:
>> >>>
>> >>> /** Get the rank from a featureprop record.
>> >>> * <featureprop>
>> >>> * <rank>0</rank>
>> >>> * <type_id>
>> >>> * Returns the rank as a short (default is 0 if it can't be parsed).
>> >>> * 12/2005: No longer used (not roundtripping featureprop ranks). */
>> >>>
>> >>> Looks like someone else was doing this same roundtrip and decided
>> >>> that this diff was not a dealbreaker. I found several other places in the
>> >>> Apollo code where rank is retrieved from Chado and assigned, but can't
>> >>> seem to find the one that is relevant here (i.e. a place where I can
>> >>> change
>> >>> the rank manually to something unique like 99999 and have that rank
>> >>> show up in the "after" XML).
>> >>>
>> >>> Any advice about why ranks aren't being pulled correctly here, and, more
>> >>> importantly, whether this is harmful? Are incorrect feature_relationship
>> >>> ranks going to hurt anything? Apollo happily loads these genes with
>> >>> the wrong ranks, and they display correctly as far as I can tell.
>> >>>
>> >>> I'd be glad to provide a chado-adapter.xml or connection info if anyone
>> >>> wants to connect to our chado and have a look first-hand.
>> >>>
>> >>> Thanks in advance,
>> >>> Justin
>> >>> _______________________________________________
>> >>> apollo mailing list
>> >>> [email protected]
>> >>> http://mail.fruitfly.org/mailman/listinfo/apollo
>> >>>
>> >>
>> >> _______________________________________________
>> >> apollo mailing list
>> >> [email protected]
>> >> http://mail.fruitfly.org/mailman/listinfo/apollo
>> >>
>> >
>> >
>>
>>
>>
>
>
--
________________
http://five.sentenc.es/
_______________________________________________
apollo mailing list
[email protected]
http://mail.fruitfly.org/mailman/listinfo/apollo
