>>November 1, 1997

 --> v31t0

version 31t of the fasta program package uses a more modular
structure for comparison functions.  In addition to modular functions
to initialize, calculate and align sequences, v31 provides a modular
function for creating the alignment display.  This was required for
fasty and fastf, which have very different alignment strategies from
the other search programs.

>>February 13, 1998

modified nascii[] so that 0, 1, 2 are no longer end of sequence
characters.

prss3 added.  Unlike prss, prss3 uses -d # to specify the number of
shuffles.

>>March 18, 1998

First public release.  Corrected problems with dropfz.c (which is
used in fasty3, tfasty3).  Makefile is well tested, but other Makefile's
are not.  PVM versions not tested.

>>March 19, 1998

Problem with unthreaded tfastx3, tfasty3 caused by bug in complib.c
fixed. All Makefiles (Makefile.alpha Makefile.sun, Makefile.sgi,
Makefile.linux) have been tested and work properly. Threaded versions
do not work on linux (yet).  Function labeling problems with fasty3,
tfasty3 corrected.

>>March 20, 1998

 --> v31t02

Fixed problem with inconsistent openlib() calls that broke BLAST databases
on some platforms.

>>March 27, 1998

 --> v31t04

Fixed a long standing problem with fastx/tfastx and fasty/tfasty that
caused various memory allocation problems and core dumps.

The PVM version works again, but cannot produce alignments.  The
change in the location of the modular display functions will require
significant changes in the pvm display functions.  For the moment,
showalign() has been commented out.

Code tested on Macintosh without changes.

Added some additional information in the results file.


Please report bugs to wrp@virginia.edu

>>April 3, 1998

Removed some debugging code in faatran.c now that fastx/fasty bugs
seem corrected.

  FASTA --> v3.14

Corrected uninitialized array elements in dropnfa.c.

>>April 10, 1998

Added facility for specifying SRCH_URL (the URL string that will be
used to re-search the database) and REF_RUL (the URL string that
will be used to lookup the sequence) ini url_subs.c.  This allows perl
scripts to provide different databases for re-searching dynamically.

>>April 16, 1998

 --> v31t05

Corrected problem with ignoring ','s in databases (','s are found in
PIR).

>>April 18, 1998

Corrected some problems with sequence names for Entrez lookups and
re-searching databases.

Made minor modifications to nxgetaa.c and compacc.c for compatibility
with Borland 'C' compiler for Win32 systems.  Including makefile.tc
fasta.rsp, prss.rsp, and test.bat for Borland 'C'/win32.

>>April 24, 1998

 --> v31t06

Fixed another bug in fasty3/tfasty3 alignment routines.

Added additional information to the do_url1() (url_subs.c) function.
The re-search URL can now reference the start, stop, and length of the
library sequence to be re-searched with.  For DNA library sequences,
these values are always in nucleotides, even with tfasta/x/y.


>>May 12, 1998

(no version change as v31t06 was not released prior to this)

Correct nxgetaa.c GETLIB to deal correctly with BLAST NR database
sequences with exceptionally long title lines.

Fix bug with long -O results files.

>>May 18, 1998

 --> v31t07

Corrected some bugs in information string lengths (e.g. gstring1,
stat_str), disabling statistics with -z 0, translation of 'X' by
saatran() (faatran.c) that caused problems with FASTX.

A serious bug has been fixed in the FASTX alignment routines.
For some pathological sequences, % identity increases from < 10%
to 40%. The version number of the main program has not changed,
but the version number of the fastx function has changed to 3.2.

>>June 19, 1998

 --> v31t08

Corrected some problems with alignments with -m 10.

Added -Z db_size option to modify apparent database size for
expectation value calculation (used only for protein/protein FASTA and
SSEARCH, FASTX, FASTY, TFASTX, and TFASTY).

>>July 1, 1998

 (no version change)

Corrected size of lbnames[], lb_size[] in structs.h to accomodate MAX_LF
files.

>>July 13, 1998

 --> v31t09

Corrected problem in nxgetaa.c encountered when reading long sequences
(that must be split) in fasta format.

Corrected problem in statistics calculation encountered with a small number
of very long DNA sequences.

>>July 17, 1998

 (no version change, date change for ssearch3)

Corrected default expectation cutoff (it was 10, now it is 2.0) for
DNA with ssearch3.