BIRCHv2.8

From Bioinformatics.Org Wiki

• If people type 'biolegato', they should now get a message telling them about bldna etc." The message should look something like this:
  • birch - general launcher for BIRCH programs (new)
  • bldna - biolegato interface for DNA programs (replaces biolegato)
  • blprotein - biolegato interface for protein programs (replaces biolegato)
  • blmarker - biolegato interface for molecular markers (replaces mgde)
  • bltree - biolegato interface for phylogenetic trees (replaces tgde)
• Make sure update-local correctly adds stuff from 2.8 to an existing BIRCH installation during update.
  • Most important thing to check is how good a job birchconfig does of copying over the local .items from GDE. Maybe its best just to try it out on bcsh01? May need to also create some GDE menus on a test system to see if they're copied properly.
  • In $birch/local/dat, we need to add a seqcolors directory to the bl* directories. Or do we? Does Biolegato look at these?
• Clean up tutorials to reflect new bioLegato instances
• Test updating of an existing BIRCH install using getbirch.
• birchhome.sh - make sure that passwd.wrm is correctly changed in both birchdb and lbirchdb during an install. (This should already be fixed in birchhome.py)
• bioLegato - when reading a file of any type using Import Foreign Format, sequences appear to be read in normally, and the names all appear. However, when the same sequences are written to a file using "Export Foreign Format", only the first sequence written to file has it's correct name. All others are given the name "nameless_x", where x is an integer.
• bioLegato - When typing in a new sequence, neither the Backspace nor the Delete key can delete the last (3'-most) nucleotide in a sequence. Looks like an array indexing error.
• bioLegato - In some menus in bldna (eg. NUMSEQ, blrevcomp, DNAPARS, DXHOM) labels are truncated. For example, "DNA" becomes "D...", in NUMSEQ and DXHOM). This occurs despite the fact that there is plenty of room in the menu. This may be a JVM error for Java 1.7
• bioLegato - In some cases, the argvalue setting on a choice_menu is ignored. This is seen with $birch/dat/bldna/menus/Similarity/GGSEARCH.item for the ZED parameter. In the same menu the EVALUE parameter is also a choice_menu, but this time argvalue is correctly used to set the default value. The same parameters are used for many of the other programs in the Similarity menus of bldna and blprotein, with similar results.
• (This may be unnecessary. ALIGN is an older program, and at first glance it looks like it is superseded by SSEARCH3. If that is true, we just get rid of ALIGN and fastaout.sh.) Need to rewrite fastaout.csh to be able to correctly send output of ALIGN to either bldna or blprotein. Probably the best approach is to pass the name of the program as a parameter to this script. Only ALIGN.itme uses fastaout.sh. (Note: fastaout.csh is obsolete. It is only used by GDE).
• bioLegato - some temporary files aren't being deleted. FEATURES leaves behind temp.xxxxx.fea. There are some others as well, which will require a bit more looking into. The answer is likely either in the scripts or the .item files.
• Groups - There seems to be a problem with groups. Try this:
  

1. in bldna, read in GMCAB2, GMCAB3, PEACAB15 and WHTCAB from $birch/tutorials/bioLegato/dotmatrix directory.
2. In name pane, SelectAll; Database --> Features - extract by FeatureKeys. This should give a new bldna window with the coding sequences.
3. Select GMCAB2, GMCAB3, PEACAB15. Group.
4. repeat 2. New bldna window pops up, but it is empty.
5. It is now impossible to Ungroup these sequences.

The problem seems to be how bioLegato exports the sequences, or rather fails to. BL creates a temporary file called temp.xxxx.fea, which is empty. It should contain fasta-formatted sequences. The empty file results in a splitdb job getting into an infinite loop.
There seems to be a separate problem. In step 1, if you never click inside the sequence pane, nothing gets selected. It's not enough to select names. At least once, you must click inside the sequence pane.

• Running external programs from bioLegato - currently, bioLegato does not run programs in the backround. As long as any child process is still running, killing the parent bioLegato window kills the child process as well. Surprisingly, even if the itemmethod ends with an amperand '&', the parent bioLegato will still persist until all child processes are complete. The reason is the java Runtime methods (rt) run system calls directly. You need to explicitly add a call to the shell to get things to run externally. For example, in $birch/install-birch/birchconfig/src/birchconfig/InstallFrame.java the following statements run the browser.failsafe.csh script:

String COMMAND = new String( "/bin/csh ../script/browser.failsafe.csh " + BP.homedir + "/public_html/index.html &");
OKAY = runner.runCommand(COMMAND);

In bioLegato, you should probably call all itemmethod's using /bin/sh.

• Methods without parameters - In previous versions (eg. 0.6.2) when there were no parameters for a program, the preferred behavior for bioLegato was to launch the program directly without creating a menu. However, this behavior has reverted to the old behavior, in which a very small menu with only a Run and Help button, will appear. Because this window is very small the user might not even notice it is there, and will think bioLegato has frozen up. The correct behavior when choosing a menu item with no parameters is to directly run the itemmethod, and not pop up any menu.
• bldna, blprotein - jalview.item no longer works! Is this something to do with how data is exported? Fix: need to run 'sleep 100' between launching Jalview and deleting temp files.
• multalign tutorial - needs to be updated to describe how to export files in the new version of Jalview.
• Fixed mrtrans - mrtrans no longer works. It looks like some strange filenames are being generated that messes up running the program. The best solution is probably to put most of the work into mrtrans.csh.
• This next one is caused when trying to use Import Foreign Format to read a FASTA file eg. $tutorials/bioLegato/multalign/defensin.CDS.fsa. This time, readseq runs, but no file is ever imported into bldna. The sequence pane remains blank.Note: this problem was caused when biolegato was changed to send a fully-qualified path to scripts that were expecting a relative filename. This is the second time this problem has been fixed.
  

Readseq version 2.1.27 (19-Apr-2008)

• Seahawk - requires Java 1.6 or higher. Solution: Pre-install instructions now require Java 1.6 or higher.
• Integrate the revised Python scripts into BIRCH and test in-situ (Dale)
• --DONE migrate birchconfig to Netbeans (Graham) - the way to do this is to make a copy of the birchconfig directory, and freeze the old one. That way, people would still install using the old one, but the new one can be distributed with every release for testing purposes, until we are satisfied with it.
• Reorganization of biolegato instances (Andrea)

• • birch - the high-level launcher for all biolegato and other GUI programs
  • bldata - the successor to dGDE. we will need to come up with a replacement for the old SeqHound web services
  • biolegato - typing 'biolegato' should prompt the user to use either bldna or blprotein

• *tutorial in overview for markers no longer works - The cause was bioLegato exporting 'in1' files as a fully-qualified path, rather than a basename. Graham has fixed this, and now all Phylip scripts work.
• create biolegato launchers for various desktops (Dale), in the order listed below. OSX will be the hardest, so we save that for last. The way this should work is that there is a central BIRCH copy of each launcher. When each user runs the newuser script (and also when they login) BIRCH checks for symbolic links to the BIRCH copy of the launcher. For GNOME, this would be something like $HOME/.gnome2/vfolders/BIRCH. Since the newuser script creates these links, the nobirch script would have to have code that deletes them.