EMBOSS: Project Meeting (Feb 21st 2000)


Sanger Centre: Peter Rice, Richard Bruskiewich
HGMP: Alan Bleasby, Gary Williams, Sinead O'Leary, Mark Faller
Apologies: Val Curwen, Ian Longden

1. Matters Arising

2. General progress on release 0.0.4

GNU sort is giving problems. The default sort order for the temporary files during indexing is different from the 'standard' sort function on other systems.

Alan reported that the system call on Linux systems is uninterruptable. These should all be replaced by exec calls (e.g. ajSystem). See also the earlier discussion of the shell on Windows systems.

Some applications are writing directly to stdout instead of using an output file.

HGMP database definitions are all working, and can be copied into the EMBOSS documentation as examples.

fuzznuc now searches both strands.

Some applications are using seqall rather than sequence when users would expect the to use single sequences. Examples include needle and water.

palindrome has been updated to limit the range of the inverted repeat. This dramatically reduces the run time for long sequences. As written, the code still searches a little beyond this range so extra hits are discarded. Code has been added to (optionally) remove overlapping shorter hits. For simple repeats these will not be required. For more complex cases they could be useful.

The -acdpretty option still runs the program. It should terminate in the same way as -help and -acdtable.

Gary would like seqret to output the entire text of an EMBL entry. Peter will investigate. This may be a new application (e.g. seqdoc) that simply dumps the entire text to a text output file.

PISE is installed at HGMP. wossname is working, but environment variables are not yet set correctly for other applications.

3. Feature Tables

GFF format does not clearly define joins across sequences. These are common in EMBL entries, but usually avoided when dumping features and sequences from ACEDB.

Several alternatives were discussed, without a clear conclusion being reached.

4. Windows Builds

Richard and Ian are looking into shell scripts (from Simon Kelley) to replace makefiles for a Windows NT build. Richard may also try a manual build.

Graphics for a Windows version may need a platform dependent module with substitute calls. There are only a few PLPLOT calls used directly from ajgraph.c

5. Any Other Business

5. Next meeting

Next meeting will be Monday 28th February 2000, 11:00am, usual place.