The third beta release of Pregap4 is now available for
download. Please see
http://www.mrc-lmb.cam.ac.uk/pubseq/pregap4.html
The changes since beta 2 follow this posting.
Please report any problems to jkb at mrc-lmb.cam.ac.uk. I'll be away with
work during most of the next fortnight and so email responses may be
delayed by a few days.
James
* Changed the default list of modules to include pretty much everything,
but with most things disabled.
* Hooked up help buttons to the online help.
* Improved handling of file of filenames. When the fofn is local we
don't add the directory component to the front of each file (this
breaks phrap, cap2 and fakii). The output prefix name now also is
the basename of the fofn, not the full fofn name.
* Allow an error of 5 bases in computing the SL and SR values in
cross_match_svec, instead of only accepting a match extending right to
the start or end of the sequence. The optimal alignment is sometimes
from (eg) 3 to 34 instead of 1 to 34, but we still want to write "SL
34". The default tag types for other matches is now SVEC, but it's
also now adjustable (tag_type local variable).
* Use init::*_use_sample_name instead of *use_sample_name in the file
conversion steps. This is because I moved these variables from global
space into 'init' space - possibly not wise! This now allows for the
"Use ABI sample names?" questions to work properly, again.
* Added a create_dialogue proc to the augment modules to point users in
the direction of Simple Text Database and Experiment File Line Types
windows.
* Added a version number to the config file. It's not used at
present, but may be useful in future for updating anything which
could break the config file format, allowing us to autodetect and
autoconvert the config file.
* Renamed the modules to .p4m and the templates to .p4t. This avoids
possible confusion with perl modules (.pm).
* Tidied up pregap4 -nowin output by adding some textual headers.
* Renamed "Select Modules" to "Add/Remove Modules" and removed it
from the tabbed note book (it's still in the menu).
* The 'cancel' command from file browsers now works correctly.
Previously it was clearing the currently specified filename.
* Make sure the filename component of a full pathname is always
visible at startup in the entry boxes.
* Don't supply blank sequences to blast as this crashes it. It's
still possible to crash blast with many other cases (all Ns, all
ambig. codes, sequences (length >= 7) entirely the same letter),
but that's blast bugs and not our fault.
* Fixed file descriptor leak in read_exp_file (used in cross_match_svec).
* Removed the vector file name query from the screen vector module. It
was misleading - it always uses the same vector that sequence vector
clip uses.
* Added a "Experiment File Line Types" configuration section to
specify things like "SI" (insert size).
* The default window style is now compact.
* Don't check error codes returned from cap2 and cap3 as they do not
return sensible values from main (no return at all).
* Check tcl errorCode to distinguish between stderr output and real
errors. Important when dealing with "sequence too short" messages,
which are just diagnostics rather than complete program failure.
This effects the screen_vector, cloning_vector_clip and
sequence_vector_clip modules.
* The "simple text database" config file generation now does "string
match" on the lines when no exact match is found. This allows for db
files like:
# ID SF SC SP
6abz91* m13mp18.vec 6249 41/-24
6aca68* puc18.vec 248 40/-28
* Added "global modules" lines to the cloning vector clip, screen
vector and trace_diff modules prevent Tk errors.
* Updated Cap3 module (and cap3_s source) to use "fofn" instead of
"outfile" as the file of filenames.
* The trev next/prev/reject/accept file buttons now check when cancel is
pressed on the save dialogue and do not move onto the next/prev file.
* The trev "-pregap_mode" mode now correctly turns off trace sequence
editing as the default mode.
--
James Bonfield (jkb at mrc-lmb.cam.ac.uk) Tel: 01223 402499 Fax: 01223 213556
Medical Research Council - Laboratory of Molecular Biology,
Hills Road, Cambridge, CB2 2QH, England.
Also see Staden Package WWW site at http://www.mrc-lmb.cam.ac.uk/pubseq/
