GFF3 exports bad coordinates about apollo

[nathandunn]

from Monica:

"- The -2147483648 coord issue - we've seen this a number of times in different instances, and it's been brought up on the WA listserv by other groups before, too. In our cases, it seems to happen for features that the annotator has deleted (or at least tried to delete)."

[cmungall]

Do we have junit tests for coords >2^32, and for all the gff-vs-interbase off-by-one edge cases?

[nathandunn]

I don’t think so. My worry from Monica’s other email is that this tends to occurs for deleted features. Most likely it is the case that a feature is deleted, so its ID is removed, but it ends up sticking around and the reference is incorrect.

I guess I could map it to a long in the code, which would fix this, but my guess is that the coordinate itself is simply wrong to begin-with. I’m still looking at it, though. If its a matter of moving everything to long’s, it shouldn’t be that bad of a fix. I have the data needed to reproduce this.

Nathan

On Oct 6, 2014, at 12:18 PM, Chris Mungall [email protected] wrote:

Do we have junit tests for coords >2^32, and for all the gff-vs-interbase off-by-one edge cases?

—
Reply to this email directly or view it on GitHub.

[cmdcolin]

I would imagine some other strange circumstance can cause these problems but I found there are some issues with 2^32 and greater. Example: 4294967296=2^32

 perl test_add_feature.pl --username admin2 --password password2 --url http://localhost:8080/apollo --featuredata '[{"location":{"fmin":2000,"fmax": 4294967296,"strand":1},"type":{"cv":{"name":"sequence"},"name":"mRNA"},"name":"TEST4","children":[{"location":{"fmin":559153,"fmax": 559153,"strand":1},"type":{"cv":{"name":"sequence"},"name":"exon"}}]}]' --track Group1.1 

Produces a GFF with end=0 and also start>end

Group1.1    .   gene    2001    0       +       Name=9D934B306620C7A4FCB6620F26FECC0C;ID=9D934B306620C7A4FCB6620F26FECC0C
Group1.1    .   mRNA    2001    0       +       Name=TEST4;Parent=9D934B306620C7A4FCB6620F26FECC0C;ID=D536CEBC603217F4FF620239583E75D8
Group1.1    .   exon    559154  559154      +       Name=D495026BDF8574A2C1E1936AB738BF61;Parent=D536CEBC603217F4FF620239583E75D8;ID=D495026BDF8574A2C1E1936AB738BF61

[nathandunn]

@deepakunni3 Can you verify that this does or does not work. Can kick to 2.0.1 if no time.

[monicacecilia]

@nathandunn shares that this is fixed and we won't see the error again on 2.0. If this resurfaces, please let us know.