DragonFly On-Line Manual Pages

Search: Section:  

RSS(1)                 DragonFly General Commands Manual                RSS(1)


NAME

       prss - test a protein sequence similarity for significance


SYNOPSIS

       prss [-Q -f # -g # -h -O file -s SMATRIX -w # ] sequence-file-1
       sequence-file-2 [ #-of-shuffles ]

       prss [-fghsw] - interactive mode


DESCRIPTION

       prss is used to evaluate the significance of a protein sequence
       similarity score by comparing two sequences and calculating optimal
       similarity scores, and then repeatedly shuffling the second sequence,
       and calculating optimal similarity scores using the Smith-Waterman
       algorithm. An extreme value distribution is then fit to the shuffled-
       sequence scores.  The characteristic parameters of the extreme value
       distribution are then used to estimate the probability that each of the
       unshuffled sequence scores would be obtained by chance in one sequence,
       or in a number of sequences equal to the number of shuffles.  This
       program is derived from rdf2, which was described by Pearson and
       Lipman, PNAS (1988) 85:2444-2448, and Pearson (Meth. Enz.  183:63-98).
       Use of the extreme value distribution for estimating the probabilities
       of similarity scores was described by Altshul and Karlin, PNAS (1990)
       87:2264-2268.  The 'z-values' calculated by rdf2 are not as informative
       as the P-values and expectations calculated by prdf.  prss uses
       calculates optimal scores using the same rigorous Smith-Waterman
       algorithm (Smith and Waterman, J. Mol. Biol. (1983) 147:195-197) used
       by the ssearch program.

       prss also allows a more sophisticated shuffling method: residues can be
       shuffled within a local window, so that the order of residues 1-10,
       11-20, etc, is destroyed but a residue in the first 10 is never swapped
       with a residue outside the first ten, and so on for each local window.


EXAMPLES

       (1)    prss  -w 10 musplfm.aa lcbo.aa

       Compare the amino acid sequence in the file musplfm.aa with that in
       lcbo.aa, then shuffle lcbo.aa 100 times using a local shuffle with a
       window of 10.  Report the significance of the unshuffled musplfm/lcbo
       comparison scores with respect to the shuffled scores.

       (2)    prss musplfm.aa lcbo.aa

       Compare the amino acid sequence in the file musplfm.aa with the
       sequences in the file lcbo.aa.

       (3)    prss

       Run prss in interactive mode.  The program will prompt for the file
       name of the two query sequence files and the number of shuffles to be
       used.  100 shuffles are calculated by default; 250 - 500 shuffles
       should provide more accurate probability estimates.


OPTIONS

       prss can be directed to change the scoring matrix, gap penalties, and
       shuffle parameters by entering options on the command line (preceeded
       by a `-'). All of the options should preceed the file names number of
       shuffles.

       -f #   Penalty for the first residue in a gap (-12 by default).

       -g #   Penalty for additional residues in a gap (-2 by default).

       -h     Do not display histogram of similarity scores.

       -Q -q  "quiet" - do not prompt for filename.

       -O filename
              send copy of results to "filename."

       -s str (SMATRIX) the filename of an alternative scoring matrix file.
              For protein sequences, BLOSUM50 is used by default; PAM250 can
              be used with the command line option -s 250(or with -s
              pam250.mat).


SEE ALSO

       ssearch(1), prdf(1), fasta(1), lfasta(1), protcodes(5)


AUTHOR

       Bill Pearson
       wrp@virginia.EDU

       The curve fitting routines in rweibull.c were provided by Phil Green,
       Washington U., St. Louis.

                                     local                              RSS(1)

Search: Section: