nda General Commands Manual nda NAME sda - to do surface density analysis of point positions on a surface SYNOPSIS sda [-comfile filename] [graph options] DESCRIPTION This program analyzes the distances between positions along a surface and computes the average density of points in objects specified as neighbor "types" around points in objects specified as reference "types". Density is computed in bins, where each bin corresponds to a range of distances, and density is the number of neighbors located at that range of distances, divided by the total area of the surface at that range of distances. The neighbor count and area are summed over all the reference points and stored internally, so that more data can be summed if desired. The input model must have one or more meshed, closed contour objects, where the mesh triangles define the surface. Points are taken from scattered point objects. Individual points sizes are not taken into account, so distances are from center to center. Each point is "dropped" onto the surface by projecting it into the plane of each tri- angle within a certain minimum distance in Z. If the projection is located within that triangle, this is a candidate position on the sur- face. The program picks the candidate position that is closest to the point as its position on the surface. The program is based on nda and nearly all the options are the same. See the nda man page for a description of the options and the gen- eral treatment of densities. Most of the initial entries match those to nda or are otherwise self-explanatory. One critical entry is for the "Minimum distance between sphere and surface in um", where the default in 30 times the Z slice thickness and the entry must be in microns. Units are in microns and the model file must have a pixel size defined, but it can be in either nm or um. Two options are different from nda: 36 to do many sets with random positions or shuffled types, and 21 to compute a kernel density over the surface and save in a model file. The latter is the origin of the surface display in "yeastpore.mod", distributed in the tiny "imod_data" sample of IMOD files. A kernel density is computed by setting up a two-dimensional "kernel" function with a certain half-width (ressem- bling a Gaussian but without tails), centering that function at each point position, and adding in the function value at each underlying point on the surface. The result is a smoothed density, where the amount of smoothing is set by an entry for the mean number of spheres to include in the kernel area. The program was developed for: Winey, M., Yarar, D., Giddings, T. H. Jr., and Mastronarde, D. N. 1997. Nuclear pore complex number and distribution throughout the Sac- charomyces cerevisiae cell cycle by three-dimensional reconstruction from electron micrographs of nuclear envelopes. Molecular Biology of the Cell 8: 2119-2132. AUTHOR David Mastronarde SEE ALSO nda, mtk BUGS Email bug reports to mast at colorado dot edu. IMOD 5.2.0 nda