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.
A month of mostly bug fixes.
We now have AceDB compiling on the RedHat 9 version of Linux which uses the newer versions of Glib and the GTK toolkit. You need to get the latest monthly release and set the ACEDB_MACHINE environment variable to "LINUX_4_RH9" to compile on RedHat 9.
Embl dumping was broken for one release because in fixing a problem with checking that #EMBL_dump_info is specified correctly I forgot to ensure that I jumped back out of the #EMBL_dump_info submodel before checking for further tags, pathetic.
Also, for worm group, I have disabled the accession number output because they have changed their model completely and the existing code simply won't work. They now put in the accession number via an external script.
blixem was crashing with some sequences/homologies and it turns out that this is because some gaps array mappings were not being clipped properly resulting in blixem attempting to access non-existent sequence data. This is now fixed.
Sometimes confirmed_introns caused acedb to crash, it turns out this was caused by the transition from fmap working at the "whole clone" level, i.e. bringing in all of any clones required to make up an fmap, to fmap clipping adjacent clones.
In this case it meant that confirmed introns of an adjoining clone were not getting mapped, this was fine until we had the situation where we had a confirmed intron being confirmed from 2 different sources. This is effectively two introns with the same coords but different confirmations. The first one did not get mapped correctly but because the second one had the same coords the code assumed it had been mapped and so tried to process the second intron as though everything was ok. Now when introns have the same coords we check to see if the previous intron was actually mapped.
Fix bug in pickRegisterClassMasks(), code assumed that a bsFlatten array only has 4 byte fields in it but this is not true on the Alpha as one of the union members is a pointer which is 8 bytes. This bug prevented acedb from initialising on the Alphas if composite classes were declared in subclasses.wrm.
Fix bug which meant that only the sequences passed in from acedb got sorted initially, the sequences obtained from another source (e.g. pfetch) did not get sorted until user selected a sort method. Now all sequences are sorted from the start.
As part of the ZMap development it is our intention to add a "ZMap" display type to acedb and to facilitate development Rob has created a new target called xacez.
This target is not compileable at the moment but will become so as we develop the ZMap display.
The main cause of this problem is two fold: the homology conversion routine was not recording any clipping of the main homology start/end coordinates by the SMapMap() call. More seriously, sMapLocalMap() does not take any notice of the start/end coords that are passed to it and does not return a mapping status to indicate any clipping.
Currently this is fixed by mapping each gap separately so we can detect if it gets mapped correctly. The better fix is to sort out sMapLocalMap() and this will be done in the next month or so.
You can pick up the monthly builds from: