Changes with 2.0  (March, 1995)

	WARNING - Optimization is now turned on by default.  The
	meaning of the "-o" option has been reversed.  "-o" now turns
	off optimization, reverting to the earlier method of sorting
	by "initn" scores.

	Change default protein matrix to BLOSUM50.  PAM250 is
	still available with -s 250.  Change program to accept gap
	penalties from the command line with "-f" (-12) and "-g" (-2).

	Provide MARKX=4, which allows one to display the conserved
	regions of the query sequence after a library search.

	Calculate explicit probability estimates for FASTA, TFASTA,
	and SSEARCH.  Estimates assume that the library contains a large
	number of unrelated sequences.  If this is not correct, the
	estimates are useless (and should be turned off with the -z
	flag).

	The width of the band used to calculate optimized scores is
	now variable.  For proteins and ktup=1, 32 residues are used,
	otherwise 16 residues are used.  For DNA, 16 residues are
	used. This value can be changed with the "-y" option.

	FASTA alignments now use the Smith-Waterman algorithm; there
	is no longer a limit on gap size for FASTA alignments. 

	Fixed a rare bug in lalign/plalign for low gap penalties.

	Fixed lfasta to read one letter filenames in second position.

April 5, 1995

	Fixed bug in blast-format file reading treat sequences that do
	not end in "*" properly.

May, 1995

	More accurate display of the expected value histogram.  The
	quality of the fit is now quantitated with the
	Kolmogorov-Smirnov statistic.

	DNA match/mismatch penalties changed to +5/-4.

	An expectation theshold (-E) is provided for displaying
	scores.

July, 1995

	Corrected a very serious bug in ssearch E()-score calculation
	for large databases.

	Corrected a minor problem with histogram scaling.

	Removed Kolmogorov-Smirnov statistic if histogram not shown.

	Show correct scoring matrix if specified matrix is not
	found.

August, 1995

	Some corrections so that "-z" flag works properly and statistical
	calculations fall back properly when no distribution of lengths
	is available.

2.0x3	Change default DNA and TFASTA alignments to older band-limited
	Smith-Waterman rather than full Smith-Waterman.  Now DNA
	sequence searches are as fast as before (with Smith-Waterman
	alignments, they were often 50 times slower).  Full Smith-Waterman
	alignments are available with the "-A" option.

	Small changes in the way that memory is allocated for
	alignments in FASTA, TFASTA, LFASTA/PLFASTA, and SSEARCH.

	The DOS/BorlandC and UNIX versions have been merged.  All
	files necessary for compilation on Dos/WinNT are included.

	Added -O option to FASTA, TFASTA, LFASTA, PRSS, PRDF, LALIGN, ALIGN
	to specify output file.

2.0u3	merge of Mac FASTA with Win/DOS, Unix FASTA to a single set of files.

Sept, 1995

	add -Q option to prss, prdf.  Fix bug in -O option for those
	programs.

	Allow longer lengths for filenames.  Use QFILE_SIZE and LFILE_SIZE
	to define lengths for query and library file names (40, 80 for
	microcomputers, 256 for Unix).

November, 1995

	Fix bug in nxgetaa.c that prevented reading multiple
	blast-formatted files.

February, 1996

	see readme.v20u4 for more information

	added -m 10 option for parseable output

	added library_type 6 for GCG formatted files

	added -L option for long descriptions of library sequences

	"randseq" random shuffle program now available.

March, 1996

	modified nxgetaa for 12 character locus names.

	fixed a bug in lfasta that appears with very long sequences

April, 1996

	Make certain '-z' option really works (required for libraries with
	sequences < 10 aa).

	Removed duplicate sw_score: in ssearch with -m 10.

	Added -DPROGRESS to report progress of search with "....".

Mar, 1996

	Added "fastx", see readme.v20u5.  "-h" is not "-H".