cds glyph producing incorrect translations

[scottcain]

In several GBrowse instances around the net (WormBase, arabidopsis, paramecium for example), the cds glyph is producing incorrect translations for second and subsequent exons in the display. That is, for whatever the first exon in order of translation that appears in the details panel (it doesn't have to be the first exon in the transcript), the translation is correct, but after that exon, the translation is incorrect. See these examples from WormBase:

Note that the incorrect translations that appear seem to come from other portions of the same transcript translated in an incorrect frame. (I determined this be transcribing the AA sequence and then doing a tblastn against the elegans genome.) So it seems that something is going wrong with coordinate mapping. Also note that in the second image, the translations in the second exons are exactly the same even though they are splicing to different places in the genome.

I tried to test the underlying BioPerl functionality by writing a "in silico" splicing and translation script, where the script pulled mRNA features from a SeqFeature::Store database and put the CDS pieces together and translated it, but that worked fine.

Live examples of where this is happening elsewhere:

Arabidopsis: http://paramecium.cgm.cnrs-gif.fr/gb2/gbrowse/ptetraurelia_mac/?start=11267;stop=11367;ref=scaffold_1;width=750;version=100;flip=0;grid=1;id=b9f0de601f367c68c264ebe77c96bae4;l=Translation%1ECDS%1EGenes

Paramecium: http://tairvm17.tacc.utexas.edu/cgi-bin/gb2/gbrowse/arabidopsis/?start=1507795;stop=1507994;ref=Chr1;width=800;version=100;flip=0;grid=1;id=0a5edc2236a0c2faa054303a7a2510af;l=FrameTranslation%1ECDS%1EProteinCoding