Figure 1
From: Genome-wide prediction, display and refinement of binding sites with information theory-based models
Architecture of the Delila-Genome system. Server programs are shown on the right side of the schema and client programs shown on the left side. A Java-based GUI application (Delgenfront) is run on a desktop client that prompts entry of a series of parameters (server, results directory, genome draft, email address) and the location of ribl file or entry of a weight matrix. These data are sent to a Linux server which runs the scan and promotsite programs to display predicted binding sites. The scan and promotsite jobs may be submitted individually or sequentially. Since scan operates on Delila books, scripts have been provided to automate the downloading and build Delila books of the genome drafts from UCSC (documented in the package: Readme.txt). The genvis program uses the results of previous chromosome or genome analyses with scan and promotsite to generate BED and HTML files of predicted binding sites within a user-defined genomic interval. Upon opening the HTML page, the user uploads the BED file to the corresponding version of the UCSC genome browser, which then displays the custom binding site track of the interval containing the site juxtaposed with other genome annotations. The HTML page is also hyperlinked to the binding site sequence (which can be used to generate a sequence walker using the autolist script), details of the binding site location, and the GenBank and SOURCE entries of the transcript associated with the site. Results obtained with different information matrices can be compared with the scandiff program, which generates BED files for binding sites found with each of the matrices and summary output indicating these differences. While promotsite takes input parameters in a file, all other Delila-Genome programs have command line options to specify the required and optional parameters and most support an '-h' switch that displays these options.