September 18, 1991
New versions of the Basic Local Alignment Search Tool (BLAST) family
of programs BLASTP, BLASTN, BLASTX, and BLAST3, along with a new addition
to the family, TBLASTN, are available for anonymous ftp on ncbi.nlm.nih.gov
(IP address 130.14.20.1). Source code for these and other programs is
posted in the pub/blast, pub/dfa, pub/gish, and pub/ncbi subdirectories.
Features of the distribution include:
o a new program TBLASTN, that compares an amino acid query sequence against
a nucleotide sequence database. By default, the database is dynamically
translated and searched in all 6 reading frames (3 on each strand), with
command line options to search just the "top" or just the "bottom" strand
and to select from one of nine built-in genetic codes.
o a new utility program named "pam" to generate PAM substitution matrices
of any desired generation (from PAM 2 to PAM 511) and scale.
o codons which contain IUPAC ambiguity letters are translated by BLASTX and
TBLASTN into specific amino acids when such inference is possible.
o second pass, low stringency (using a lower cutoff score) comparisons
against a database sequence are performed by BLASTP and TBLASTN when they
find at least one alignment against a sequence that satisfies the primary
cutoff score.
o improved report format, including: a summary section of one-line
sequence descriptions of those database sequences whose alignments
satisfied the cutoff score; and a histogram of maximum alignment
scores for all sequences in the database.
o each of the 3 sections of the output report (histogram, one-line summaries,
and sequence alignments) can be independently turned off or regulated
by command line options.
o BLASTDB and BLASTPAM environment variables can be used to point the
programs to alternate directories for database and substitution
matrix files.
o tested compatibility with several UNIX platforms, including: Sun
Microsystems SunOS 4.1.1, Silicon Graphics IRIX 3.3.2, Next 2.1,
Apple A/UX 2.0.1, AIX/370 1.2.0; at least partial compatibility
with Hewlett-Packard HP/UX and Cray Research Unicos.
o support for multiprocessing on Silicon Graphics (SGI) PowerIRIS computers
running SGI IRIX System V Release 3.3.2 or later, providing typical
overall improvement in speed of about 7-fold on machines with 8 processors.
o a new default word length (W) of 3 in the programs BLASTP, BLASTX, TBLASTN,
and BLAST3, which reduces memory demands at a modest cost in overall
execution speed.
o rather than using fixed constants, the default values for parameters
T (neighborhood word score threshold) and X (alignment extension drop-off
score) are calculated dynamically, using ad hoc equations that utilize
measures of information for each individual database query.
o the default, calculated value for the neighborhood word score threshold
T is now lower than in the previous distribution, to obtain high
sensitivity by default (at the expense of increased execution time).
o on systems which support it, improved utilities to manage memory-resident
database files using UNIX System V interprocess communication facilities.
Warren Gish phone: (301) 496-2475, ext. 64
Staff Fellow FAX: (301) 480-9241
National Center Internet: gish at ncbi.nlm.nih.gov
for Biotechnology Information
National Library of Medicine
Building 38A, Room 8S-806
8600 Rockville Pike
Bethesda, MD 20894-0001
