Coordinate mismatch for extremely small exon regions

[nsalomonis]

Description: A novel splice junction was observed for the human EnsMart91 database from an RNA-Seq analysis, where the assigned junction IDs are incorrect (relative to the reported coordinates, which are correct).

AGAP5:ENSG00000172650:I2.1_73696925-E14.1|ENSG00000172650:E2.1-E3.1 with the coordinates
chr10:73696925-73694804|chr10:73697097-73694806

The exon coordinates in this region are:
ENSG00000172650 E3.1 chr10 - 73694804 73694806
ENSG00000172650 E3.2 chr10 - 73694736 73694804

The problem is complicated by the fact that E14.1 doesn't exist, and true 3' exon is E3.1 not E14.1 and that E3.2 starts at 73694804 and not 73694806. Hence this is possibly an issue with extremely small exon regions in the database build process (EnsemblImport.py) and the RNA-Seq annotation alignment function (RNASeq.py - annotateNovelJunctions() ) .

Evaluation: There are two problems here:

1. Database coordinate reference issue (database build)
2. Splice-junction annotation (bed file import)

For issue 1: Rebuild the database for just this chromosome region and determine where the bug is trigged and under what circumstances (does not appear to be a wide-spread issue).
For issue 2: Once the above is complete, verify the issue still persists with a sample that contains reads aligning to the novel junction for selected samples (AML). If so, determine why E14.1 is being assigned (is it from the prior gene model, left over - does it break the model).