NAME

       hmm_tweak - Alter transition probabilities in an HMM definition file.

DESCRIPTION

       Alter  transition probabilities in an HMM definition file.  After specified operations are
       performed, transition probabilities are renormalized and the adjusted file is  written  to
       standard out.  This program may be used multiple times in a pipe.

USAGE

       hmm_tweak [OPTIONS] <file.hmm> <cmap.cm>

OPTIONS

       -f <cats>
              Operate  on transitions *from* states corresponding to the specified category names
              (default all) -t <cats> Operate on transitions *to*  states  corresponding  to  the
              specified  category names (default all) -m <fact> Multiply transition probabilities
              by the specified factor.

       -a <const>  Add  the  specified  constant  to  transition  probabilities.   -e  <val>  Set
              transition probabilities equal to the specified value.

       -i <icats> Assume a phylo-HMM indel model for states corresponding to

              the  specified  category  names.   -u  <tree>  (Required with -i) Assume given tree
              topology (.nh file).  -F <gps> (For use with -i) Operate on transitions from states
              corresp.  to the specified gap-pattern numbers (ANDed with -f).

       -T <gps>
              (For  use  with  -i)  Operate  on  transitions to states corresp.  to the specified
              gap-pattern numbers (ANDed with -t).  -z Equalize  transition  probabilities.   Set
              all  transition  probabilities  indicated  by  -f/-t/-F/-T to their overall average
              value.  Options -m and/or -a can be used to adjust this average value.  -R Restrict
              to  successive  transitions  within  a  category range.  -y Like -z, except compute
              separate averages for five classes of transitions, based on the gap patterns of the
              states  involved:  between  null gap patterns, between equal non-null gap patterns,
              from null to non-null gap patterns, from non-null to null  gap  patterns,  and  all
              others.   Useful  with  the  indel  model  when  training data is sparse (e.g., for
              splice-site states).  Options -m and -a will be applied to transitions of  the  3rd
              and 5th classes described.

       -h     Print this help message.