Provided by: pftools_3+dfsg-3_amd64 
NAME
pfmake - generate a profile from a multiple sequence alignment
SYNOPSIS
pfmake [ -0123abes ] [ msf-file | - ] score-matrix [ profile-file ]
[E=#] [F=#] [G=#] [H=#] [I=#] [L=#] [M=#] [S=#] [T=#] [X=#]
DESCRIPTION
pfmake generates a PROSITE profile from a multiple sequence alignment using methods
described by Gribskov et al. (1990), Luethy et al. (1994), and Thompson et al. (1994),
with modifications to exploit the features of the new profile format. The file containing
the multiple sequence alignment (msf-file) must be in MSF format as generated by GCG
programs or by readseq (checksums are ignored). The score-matrix file must also be in GCG
format. If `-' is specified instead of a real filename, the multiple sequence alignment
is read from the standard input.
If an already existing profile is given as input via the third optional argument, the
parameters of the DISJOINT, NORMALIZATION, AND CUT_OFF blocks will be read from input, all
other profile parameters will be recalculated. Header and footer lines outside the matrix
block will also be transferred from input to output.
If no input profile is given, the disjointness definition will be set to PROTECT with
borders leaving short unprotected tails (maximum 5 positions) at the beginning and at the
end of the profile. Furthermore, one normalization mode (n-score = raw-score / F , where F
is the output score multiplier , see below), and two cut-off values (level 0:8.5, level
-1:6.5) will be defined.
OPTIONS
-0 Global alignment mode; initiation (termination) at low cost is possible only if the
alignment starts at the beginning (end) of the profile and at the beginning (end)
of the sequence.
-1 Domain global alignment mode; initiation (termination) at low cost is possible only
at the beginning (end) of the profile; it may start and end at any position within
the sequence.
-2 Semi-global alignment mode; initiation (termination) at low cost is possible if the
alignment starts either at the beginning (end) of the profile or at the beginning
(end) of the sequences. This is the default alignment mode.
-3 Local alignment mode; initiation (termination) at low cost is possible anywhere.
The high-cost initiation/termination score (parameter H) is meaningless.
-a Causes pfsearch to weight gaps asymmetrically, as in Gribskov et al. (1990).
-b Block profile mode. By imposing additional constraints on the placement of
insertions and deletions, this mode produces profiles that favor alignments with
insertions and deletions positioned symmetrically around a few positions. For each
gap region a gap center is defined which usually corresponds to the place where gap
excision has been applied (see parameter X). If no gap excision has been applied,
the position is chosen such as to maximize the sum of deletion opening events
before, and deletion closing events after the gap center. Within a given gap
region reduced deletion opening penalties are offered only before, reduced deletion
closing penalties only after, and reduced insertion penalties only at the center.
This option is incompatible with options -a and -e and automatically disables them.
-c Circular profile. The topology of the profile is declared as circular. The first
and the last insert positions are merged by retaining the higher value of each
parameter type.
-e Enables endgap-weighting mode as implemented in the GCG program ProfileMake.
Endgaps in the multiple sequence alignment will be interpreted as deletions
relative to the other sequences and thus be considered for the delineation of gap
regions. The default is no endgap weighting as introduced by Thompson et al.
(1994) in the program ProfileWeight.
-s Causes pfsearch to weight gaps symmetrically (default mode). The initial gap
opening scores (MD, MI) computed from the maximal gap length and the command-line
parameters E,G,I, and M, will be divided by two and the resulting value will be
assigned to both gap opening and gap closing scores (MI, IM, MD, DM).
PARAMETERS
E=# Gap extension penalty, see Gribskov et al. (1990). Default: E=0.2 (appropriate for
1/3 bit-scaled blosum45 matrix).
F=# Output score multiplier. On output, all profile scores are multiplied by this
factor and rounded to nearest integers. Default: F=100.
G=# Gap opening penalty, see Gribskov et al. (1990). Default: G=2.1 (appropriate for
1/3 bit-scaled blosum45 matrix).
H=# High-cost initiation/termination score. This score will be applied to all external
and internal initiation and termination scores corresponding to path matrix
positions where initiation or termination at low cost is not possible according to
the alignment mode specified. Default: H=* (low-value).
I=# Gap penalty multiplier increment, see Gribskov et al. (1990). Default: I=0.1.
L=# Low-cost initiation/termination score. This score will be applied to all external
and internal initiation and termination scores corresponding to path matrix
positions where initiation or termination at low cost is possible according to the
alignment mode specified. Default: L=0.
M=# Maximum gap penalty multiplier, see Gribskov et al. (1990). Default: M=0.333.
S=# Score matrix multiplier. On input, the numbers of the score matrix are multiplied
by this factor. Default: S=0.1.
T=# Gap region threshold. This is the minimal fraction of gap characters a column of
the multiple sequence alignment must contain in order to be considered part of a
gap region. Default: T=0.01.
X=# Gap excision threshold. This is the minimal fraction of non-gap characters a column
of the multiple sequence alignment must contain in order to be converted into a
match position. The IM and MI transition scores of insert positions corresponding
to excised columns are set to zero; the other parameters remain unchanged.
Default: X=0.5.
EXAMPLES
(1) pfmake -b1 sh3.msf blosum45.cmp H=0.6 > sh3_block.prf
Generates a domain-global block profile from a multiple alignment of SH3 domains
using the blosum45 matrix. sh3.msf contains a multiple alignment of 20 SH3 domains
from SWISS-PROT release 32 including sequence weights. blosum45.cmp contains a 1/3
bits-scaled blosum45 matrix in GCG format. Note that fragment matches (alignments
to parts of the profile) are not prohibited but penalized by the parameter H=0.6.
REFERENCES
Bucher P, Karplus K, Moeri N & Hofmann, K. (1996). A flexible motif search technique
based on generalized profiles. Comput. Chem. 20:3-24.
Gribskov M, Luethy R & Eisenberg D (1990). Profile analysis. Meth. Enzymol.
183:146-159.
Luethy R, Xenarios I & Bucher P (1994). Improving the sensitivity of the sequence profile
method. Prot. Sci. 3:139-146.
Thompson JD, Higgins DG & Gibson TJ (1994) Improved sensitivity of profile searches
through the use of sequence weights and gap excision. Comput. Appl. Biosci. 10:19-29.
AUTHOR
Philipp Bucher
Philipp.Bucher@isrec.unil.ch