|
|
EMBOSS: Project Meeting (Mon 19th February 2007) |
Peter commented on a request for an application that returns the sequences from a restriction digest. It is not possible to do this with report output, because we have no way to support a listfile (USA) output for a join across the ends of a circular sequence. The simple solution is to write an application that produces sequence output, either directly or using -rformat fasta.
A very useful warning from the Microsoft compiler was where a unary minus is applied to an unsigned value. No other compiler reported any problem.
Alan is now reading up on X3D for the graphics library code.
Peter has updated the handling of pattern files (Henrikki's contribution that finally was added in EMBOSS 4.0.0). The pattern names are now used (the -pname associated qualifier) where there is no pattern name in the input file. Two pattern formats are recognized, supported by the -pformat associated qualifier. For now, it is easy to automatically detect the format: one starts with '>' in FASTA style, the other is just the pattern text.
Mahmut has been working on generating Feta descriptions for myGrid advanced searches of application metadata. These are working, and will be used by Franck Tanoh to update his descriptions in Manchester.
SoapLab had a bug in the handling of default true values being set to false. This was fixed, and the AppLab server restarted.
SoapLab logfiles have been checked for errors. Most are the use of wrong database names or sequence identifiers that were not found. The database information page will be updated to guide users.
SoapLab services for EMBRACE have been very stable over the past month.
Shaun is implementing a despatcher to submit jobs to run EMBOSS binaries. This will require generation of valid commandlines using the Apache annotations library and JAX-DB bindings. A Maven plugin will be used to annotate as a first stage in the software build. Logging and protection against bad input strings are required. A prototype system will be shown at next month's EMBRACE workshop in Uppsala.
Peter has added all the QA tests to valgrind.pl. Some QA tests have special handling of standard input because the programs are interactive. These include the MSE editor, the EMNU menu system, HMM calibration, and phylip utilities for tree editing and drawing. If there is a file available, it is simply copied. In other cases a file is provided based on part of the QA test's standard input.
The input files are copied in a preprocessing step defined for each named test. If the QA tests are renumbered, the automatic *-qa1, *-qa2 naming will break but a simple edit of valgrind.pl will match the inputs to the tests again.
There has to be at least one QA test for each application, and now also one memory test through this mechanism. We still need to add QA tests for major program options so that all commandline qualifiers are tested.
We can make a release 4.1.0 in early March. Peter will confirm when the QA and memory tests all pass.
Jon has updated the Admin chapter - a lot of markup is still needed. Word files have been converted to DocBook using a utility with a free trial period. The results will be committed to CVS next later in the week. Most of the required markup is "screen" to highlight on-screen elements, and system items to be displayed in a different font.
Jon gave a talk to EBI pre-docs and a demonstration. The developers tutorial was left as an exercise for those interested in coding. There will be a second session in May which can include an overview of EMBRACE.
Jon reported on the status of Mike Hurley's code for structure alignment using a dynamic programming two-step algorithm comparing patterns of physical residue-residue contacts.
The algorithm runs relatively slowly and has only been used on small use cases. It is a possible candidate for testing EMBOSS on high performance computers or grids.
The next meeting is on Monday 5th March.