If you want to have this newsletter mailed to you or you want to make comments/suggestions about the format/content then send an email to acedb@sanger.ac.uk.
This month sees better support for Confirmed Introns in FMap, some changes to EMBL dumping to better support SMap'd classes, some improvements to GFF output and yes, the usual bug fixes.
If you click on a Confirmed Intron in FMap you will now see which sequence(s) confirms that intron in the blue report line at the top of FMap.
If you don't use the EMBL dumping options from AceDB then you don't really need to read this.
Changes to the wormbase database models have enforced some changes to the EMBL dumping routines in AceDB, this will almost certainly affect you if you use the AceDB EMBL dump routines.
Only objects that specify an EMBL_dump_method via the EMBL_dump_XXXX tags will now get dumped, otherwise they will be ignored, this is a tightening up on previous behaviour.
The code now knows how to handle the new Wormbase CDS_child, Transcript, Pseudogene and Transposon classes. The Pseudogene tag has changed to either Coding_pseudogene or RNA_pseudogene. This has required small but significant changes to the code that handles CDS vs. RNA and whether something is a pseudogene. If you make use of these tags you should try out the EMBL dump routines before submitting data to EMBL.
We have added a new FEATURE_OBJ seg type for more generalised features that result from smap objects that are not Sequence class objects. These new segs can be handled in a more generalised way than before.
GFF dumping has been improved so that parent/child related features are dumped properly, i.e. a CDS with is relevant exons/introns is now labelled correctly. This includes better support for the GFF_XXX tags in the Method class.
A stupid bug in setting the position of confirmed introns meant that in certain circumstances they could be off by one on the forward strand and off by two on the reverse. The basic cause of the bug was forgetting that database coords are 1-based but FMap segs are 0-based.
Fixing a bug in blixem whereby % identity was not correctly calculated for gapped alignments revealed a bug in sMapLocalMapEx() which caused the nucleic acid coords for a protein alignment to be truncated.
Fixed a number of problems with blixem:
December has seen much work to get our systems into a state where we can get on with the detailed implementation of ZMap, we've decided to use more of the Gnome/GTK software as it is widely ported and most readily available on Linux.
We've decide for instance to make use the Gnome Canvas for drawing the sequence etc. as this provides many useful features including complete support for event handling based on what in the Graph package would be "boxes".
Added classModel2Class() function that given the key that equates to "?>model<" (e.g. "?Sequence"), as found in the model for a class, returns the int that represents that class, returns 0 on failure. This slightly odd looking function is useful when writing code that needs to trawl through the models for a number of classes, see code in w7/fmapgene.c which needs to dynamically examine the database models and record which classes contain certain tags.
Rob Clack has been making more changes to our build/test system to further improve its automation and save us work in the future. He has also fixed a number of problems as Sanger systems have migrated to new levels of Linux, OSF etc.
You can pick up the monthly builds from: