will get unique regions with variation_feature, stops it from counting snps twice on chromosome Y

Pushing pig specific hacks for the naming of transcripts and the overwriting of xref pipeline assigned values (in a similar way we do for zfish)

Fix for parsing files from Ken @ CSHL. ORIGIN was used in another context apart from notifying the start of the sequence in a genbank a.thaliana file

for gene display xref, ignore gene xrefs where uniprot_genename is dependent on a low-evidence uniprot entry

do not use uniprot_genename as display xref if the master xref it is dependent on is of low evidence

The old way was possibly the WORST way to blank a LRU cache by replacing it for a bare hash. Possible memory leak fixed ...

rather than return all toplevel slices, returns all slices with genes
does not affect much species with a reduced number of toplevel slices but increases things considerably for large number of toplevel slices

Adding patch from James Allen where a copy/paste bug has meant the incorrect setting of end_Exon to be on the $translation and not alternative $translation

also supports trinomial species names

Imports recombination maps.

when counting features, do not use strict lower/higher