DNA/GUI works with single-marker gels. There must be a least two marker lanes, preferably to the left and right of other lanes. A centered third marker lane can be useful. There must be at least four bands in each marker lane, and the top and bottom bands' vertical coordinates should bracket the bands in the experimental lanes. Images may be obtained from directly fluoresced gels or from transilluminated autoradiograms.
Digitized images may be of any dimensions up to 1024 x 1024. If the image is greater than 512 pixels in either direction, it is downsampled or halved in each direction in order to fit the 512 x 512 space allocated for it. Also, if the image is larger than the display space, the original image is retained in memory for analysis. An image with dimensions no greater than 512 x 512 is called a small image while one with greater dimensions is called a large image. The program does not currently handle images larger than 1024 x 1024.
DNA/GUI works with images which are in a local (Washington University)
import format. Two other kinds of images may be input
(see following two sections), but these are converted to the
import format for analysis.
The import format is a simple concept; an
imort image is represented as two files, the image binary
data file and an ascii import file of tags describing the image. In the data
file, image data are stored in binary in row-major order.
The ascii import file of descriptors includes image size, type of
binary image data, and all details about the analysis of the image.
The descriptors may be found in
Appendix A.6.
An exception to the two-file concept is the individual lane image
for which the image data are embedded in the ascii import file.
Lane images are created, as an option, when the results of analysis
are saved.
The ascii import file typically has the same name as the data file,
except that the import file is suffixed ``.imp''.
File naming conventions are described in
Section 2.9.
When accessing an import image, one specifies the ascii
import file; this, in turn, contains a string specifying the
name of the image data file.
Several local projects using DNA/GUI acquire images from Photometrics CCD camera systems. And DNA/GUI will recognize certain image types generated from these kinds of cameras. These images are, however, converted to the import format for analysis; the original images are not destroyed.
TIFF images may be input, although extensive testing with different TIFF headers has not be tried. An input TIFF image is converted to the import format (the original image is not destroyed), and the import image is used throughout the analysis. There are no immediate plans to develop TIFF output images. TIFF images dimensioned greater than 512 are proportionally downsampled to 512 in the larger dimension. The downsampling technique is bilinear interpolation. Input TIFF images should be suffixed .tif or .tiff.
If the input image is of neither of the above formats, the program will allow files with headers of known sizes (including headerless). Such images are called "odd" images. For these, the user is prompted for header size and for other image characteristics such as dimensions and data type. The header information is ignored, and the image data are converted to the import format. The original file is not destroyed. At present, the program does not deal with specific formats such as PNM, GIF, etc.