1
<TITLE>BLAST search main parameters</TITLE>
2
<!-- Changed by: Sergei Shavirin, 2-Apr-1996 -->
3
<BODY bgcolor="FFFFFF" link="0000FF" vlink="ff0000" text="000000" >
4
<h1>BLAST Search main parameters</h1>
8
<dt><b><a name = histogram>HISTOGRAM</a></b>
9
<dd>Display a histogram of scores for each search; default
10
is yes. (See parameter H in the BLAST Manual).
11
<dt><b><a name = descriptions>DESCRIPTIONS</a></b>
12
<dd>Restricts the number of short descriptions of matching
13
sequences reported to the number specified; default
14
limit is 100 descriptions. (See parameter V in the
15
manual page). See also EXPECT and CUTOFF.
16
<dt><b><a name = alignments>ALIGNMENTS</a></b>
17
<dd>Restricts database sequences to the number specified for
18
which high-scoring segment pairs (HSPs) are reported;
19
the default limit is 50. If more database sequences
20
than this happen to satisfy the statistical
21
significance threshold for reporting (see EXPECT and
22
CUTOFF below), only the matches ascribed the greatest
23
statistical significance are reported.
24
(See parameter B in the BLAST Manual).
25
<dt><b><a name = expect>EXPECT</a></b>
27
The statistical significance threshold for reporting
28
matches against database sequences; the default value
29
is 10, such that 10 matches are expected to be found
30
merely by chance, according to the stochastic model
31
of Karlin and Altschul (1990). If the statistical
32
significance ascribed to a match is greater than the
33
EXPECT threshold, the match will not be reported.
34
Lower EXPECT thresholds are more stringent, leading
35
to fewer chance matches being reported. Fractional
36
values are acceptable. (See parameter E in the BLAST
38
<dt><b><a name = cutoff>CUTOFF</a></b>
40
Cutoff score for reporting high-scoring segment pairs.
41
The default value is calculated from the EXPECT value
42
(see above). HSPs are reported for a database sequence
43
only if the statistical significance ascribed to them
44
is at least as high as would be ascribed to a lone
45
HSP having a score equal to the CUTOFF value. Higher
46
CUTOFF values are more stringent, leading to fewer
47
chance matches being reported. (See parameter S in
48
the BLAST Manual). Typically, significance thresholds
49
can be more intuitively managed using EXPECT.
50
<dt><b><a name = matrix>MATRIX</a></b>
52
Specify an alternate scoring matrix for BLASTP, BLASTX,
53
TBLASTN and TBLASTX. The default matrix is BLOSUM62
54
(Henikoff & Henikoff, 1992). The valid alternative
55
choices include: PAM40, PAM120, PAM250 and IDENTITY.
56
No alternate scoring matrices are available for BLASTN;
57
specifying the MATRIX directive in BLASTN requests
58
returns an error response.
59
<dt><b><a name = strand>STRAND</a></b>
61
Restrict a TBLASTN search to just the top or bottom
62
strand of the database sequences; or restrict a BLASTN,
63
BLASTX or TBLASTX search to just reading frames on the
64
top or bottom strand of the query sequence.
65
<dt><b><a name = filter>FILTER</a></b>
67
Mask off segments of the query sequence that have
68
low compositional complexity, as determined by the
69
SEG program of Wootton & Federhen (Computers and
70
Chemistry, 1993), or segments consisting of
71
short-periodicity internal repeats, as determined
72
by the XNU program of Claverie & States (Computers
73
and Chemistry, 1993), or, for BLASTN, by the DUST
74
program of Tatusov and Lipman (in preparation).
75
Filtering can eliminate statistically significant but
76
biologically uninteresting reports from the blast
77
output (e.g., hits against common acidic-, basic- or
78
proline-rich regions), leaving the more biologically
79
interesting regions of the query sequence available
80
for specific matching against database sequences. <BR>
82
Filtering is only applied to the query sequence (or
83
its translation products), not to database sequences.
84
Default filtering is DUST for BLASTN, SEG for other
87
It is not unusual for nothing at all to be masked
88
by SEG, XNU, or both, when applied to sequences
89
in SWISS-PROT, so filtering should not be expected to
90
always yield an effect. Furthermore, in some cases,
91
sequences are masked in their entirety, indicating that
92
the statistical significance of any matches reported
93
against the unfiltered query sequence should be suspect.
95
<dt><b><a name = ncbi-gi>NCBI-gi<a></b>
97
Causes NCBI gi identifiers to be shown in the output,
98
in addition to the accession and/or locus name.