selfalign(1)                                                      selfalign(1)

       selfalign - Realign model based on fits to trajectories

       selfalign  options  root_name|input_file  output_file

       selfalign will refine the alignment of a model of fibers that progress
       through the Z-dimension, such as microtubules modeled through serial
       sections, using the trajectories in the model.  This provides a way of
       smoothing a model without disrupting the spacings between fibers in a
       section, which would happen if trajectories were smoothed independently
       with mtsmooth.  selfalign works by first running xfmodel to
       obtain transformations relating each section to the previous one (f
       transforms).  The kind of transformation can be specified with the -t
       option, e.g., full linear, rotations and translations only, etc.  If
       there are few fibers, it is appropriate to reduce the complexity of
       transformations being solved for.  In addition, one can specify the
       range of sections in the model over which to solve for transformations,
       and thus exclude regions near the ends of the model that have few
       fibers.  If sections are excluded at the start or end of the model,
       selfalign supplies transforms for those sections by replicating the
       first or last transform actually solved for, respectively.  This pre-
       vents artifacts when local fits are made to the trends in the trans-
       forms.  The f transforms are converted to g transforms using xftoxg,
       using fitting parameters that can be adjusted with the -l and -o
       options.  The self-aligned model is produced by running imodtrans.

       The program can be run with either one or two arguments after the
       options.  Use one argument to specify a root filename.  The program
       then assumes that the input model is root_name.mod and generates an
       output model named root_name.selfmod.  Alternatively, use two arguments
       to specify the full names of the input and output models.  In this
       case, the root name for the other files is assumed to be the name of
       the input model with any extension removed.

       Run selfalign with no arguments to get a usage statement listing the
       default values of the parameters.

       -s list
              Solve for transformations over the range of sections given in
              list.  A single range (e.g., 3-32) will give best results.  The
              default is to solve over the whole extent of the model.

       -l extent
              Do local fits to transformation parameters over the number of
              sections given in extent.  This parameter is used in running

       -o order
              Set the polynomial order for the local fits to order.  This
              parameter is also used in running xftoxg.

       -t type
              Specify the type of transformations to solve for; set type to 0
              for general linear transforms, 2 for translations only, 3 for
              rotations and translations, or 4 for magnifications, rotations
              and translations.

       -i file
              Specify the name of the image stack upon which the model was
              built.  If no file is entered, the default is the root name of
              the model with extension .ali.  If such an image file is not
              available, the program will extract the needed information about
              image dimensions from the header of the model, and assume that
              this will adequately represent the information that could be
              extracted from the image file header.

       In addition to the self-aligned model, the program creates files with
       the f transforms (root_name.selfxf) and the g transforms
       (root_name.selfxg).  If any of the output files already exists a backup
       of the original is created with the ~ extension.

       David Mastronarde  <mast at colorado dot edu>

       xfmodel, xftoxg, imodtrans

BL3DEMC                              4.7.3                        selfalign(1)