NAME

       ptoh - converts a PROSITE-formatted profile into an approximately equivalent HMM

SYNOPSIS

       ptoh [ -fFhs ] [ profile-file | - ] [ random-model-file ] [L=#]
               [I=#] [D=#]

DESCRIPTION

       ptoh  converts  a  PROSITE-formatted  generalized profile into an approximately equivalent
       HMMER1- or SAM-formatted hidden Markov model (HMM).  profile-file is  a  nucleic  acid  or
       protein  profile in PROSITE format.  random-model-file contains a simple null-model in the
       format used by the HMM search programs of the HMMER1 package.  The result  is  written  to
       the standard output.

       The  profile-to-HMM  conversion  method is described in Bucher et al. (1996).  All profile
       scores corresponding to transitions which are not supported by the  HMM  architectures  of
       the  SAM and HMMER1 packages, are ignored.  If no random model is specified on the command
       line, the null model given in the GENERAL_SPEC data block of the profile is used.  If this
       is also missing, a flat residue distribution is assumed.

OPTIONS

       -f     Generate  free  insert-modules  (FIMs) at the beginning and at the end of the model
              which approximately emulate domain-global or semi-global alignment mode (see Bucher
              et al. 1996).

       -F     Generate  free  insert-modules  (FIMs) at the beginning and at the end of the model
              which approximately emulate local alignment mode.

       -h     Output in HMMER1 ascii format. This is the default.

       -s     Output in SAM format.

ADDITIONAL PARAMETERS

       D=#    Delete-to-delete transition probabilities used in conjunction  with  FIMs  (options
              -f,  -F).   This  value  will be imposed on all delete-to-delete transitions of the
              model.  The other outgoing transitions from the  affected  delete  states  will  be
              renormalized to satisfy the "sum to one" requirement.  Note that the application of
              the D parameter precedes the application  of  the  I  parameter.   Delete-to-delete
              probabilities at the beginning and at the end of the model may thus change again as
              a result  of  renormalizations  following  the  application  of  the  I  parameter.
              Default: D=0.9.

       I=#    Insert-to-insert  transition  probabilities used in the FIMs ( see options -f, -F).
              This value will be imposed on the first and the last match-to-insert and insert-to-
              insert  transition  probabilities  of the model.  The other outgoing transitions of
              the affected states will be renormalized such as to satisfy the  "summing  to  one"
              requirement.   If the -F option is simultaneously set, the I parameter will also be
              imposed on the insert-to-delete transition probability at the beginning, and on the
              delete-to-insert transition probability at the end of the model.  Default: I=0.99.

       L=#    Logarithmic  base  used for conversion of the profile scores into HMM probabilities
              (The profile scores are interpreted as log-odds ratios, see Bucher  et  al.  1996).
              This  value  over-writes  the  logarithmic base specified within the profile by the
              optional LOG_BASE parameter.  Default: 1.0233739 (1/30 bit units).

EXAMPLES

       (1)    ptoh -Fs sh3.prf standard.random L=1.0233739 > sh3.hmm

              Generates a SAM-formatted HMM for the SH3 domain from a profile.  sh3.prf  contains
              the source profile in PROSITE format.  standard.random contains the null-model used
              for conversion of the log-odds ratios into  probabilities.   The  logarithmic  base
              1.0233739  corresponds  to  1/30  bits,  the implicit scale of the profile.  The -F
              option causes the HMM to emulate local alignment (fragment search) mode.

REFERENCES

       Bucher P, Karplus K, Moeri N & Hofmann, K. (1996).   A  flexible  motif  search  technique
       based on generalized profiles.  Comput. Chem.  20:3-24.

AUTHOR

       Philipp Bucher
       Philipp.Bucher@isrec.unil.ch