|
|
EMBOSS: Project Meeting (Mon 21st September 09) |
Peter has applied the source code patches in the latest release of Phylip (3.69).
There is a proposal within EMBnet to rewrite distmat. part of the reason is that it is reported to run slowly. Peter will look into possible performance improvements.
Peter has committed further graph library refactoring code changes and the bugfixes reported in recent meetings.
Mahmut has added a memmove step to the sequence copy constructor as the quality scores were not being copied to the new object.
Alan has completed the first edit of Michael Schuster's ensembl code and commited the source files to a new ajax/ensembl directory which will appear with "cvs update -dP". There are also ajsql source files which interact directly with SQL remote servers through the libmysql-client library, and possibly also postgres if this is installed without mysql. These keep objects in memory using AjVoid and AjLong structures with reference counts.
Alan has also split out the pcre library into a separate ajax/pcre subdirectory which is used to build an epcre library.
Alan proposed changes to the configuration of the AJAX library to allow the ensembl code to be built separately. As this code needs to call AJAX functions, and to be called by ajSeqRead, we also need to split AJAX into core, sequence and ACD functions. With ACD split out, the ajcall functions should become obsolete. They allowed ajacd functions to libajaxg despite the link precedence. With the new libaries ajacd can be linked after libajaxg.
The proposed precendence for linking is:
plplot pcre ajax ajaxg ensembl ajaxseq acd nucleus application
For a cleaner structure, it may be easiest to move the acd, sequence and core AJAX functions to separate subdirectories. Alan will try to relocate the CVS server files to maintain the history of the moved files. This will mean that we cannot reproduce past releases from CVS but this is something we never attempt - past releases are available as tarballs which we can patch.
Peter is looking into the sam and bam short read formats.
Mahmut is looking into short read adaptor matching algorithms and string patterns.
Jon has added relations attributes to inputs and outputs in all ACD files. Some validation is still neeed. A new application acdrelation will help with managing EDAM relations in ACD as a "make check" application (i.e. a utility not for general users)
For the ensembl code, two aclocal files mysql.m4 and postgres.m4 include copyleft and may need some modifications (e.g. for Fedora configuration files reports a test compilation may be needed)
Peter will collect short read format data in a common location for testing.
Peter has been promised that the workstations will finally be delivered. They have now left the factory and are in the Dell system - apparently in Poland.
Peter will check why the cables, backup disk and mice ordered from Scan have not arrived yet.
It would be helpful to associate the books with release 6.2.0 to match the extended source tree and the refactored function names.
Jon will document the general principles underlying EDAM, for example meanings of relations and the terms to which they apply.
Some partners may convert to OWL. We could maintain a read-only OWL version for them. The OBO file will remain the master copy.
Peter will check on the procedures for linking EDAM with the OBO foundry when we are ready for a public release.
Peter spell checked the current EDAM file.
Jon is looking for new terms, synonyms and additional text for EDAM. One possible source is BioMOBY. Another possible source is MIRIAM from EBI.
The workshop did not define a URI string nor a syntax for annotating in SAWSDL style. We assume a GO-style website will be easy to set up once we become part of OBO.