Figure 1

Architecture and analysis flow of ColoWeb. A pre-built library of cell line specific genomic features split by chromosome is automatically built using data pulled from the UCSC Genome browser. Users submit a region of interest file via their browser to our Tomcat server. An analysis thread is started for each user request. The thread analyses a series of genomic features one at a time. Binned feature arrays tuned to the size of the analysis window are used to efficiently construct a density matrix of feature totals relative to the center of submitted regions (center matrix column). Column totals from the density matrix are normalized and summarized into a histogram. Colocation statistics including % of regions colocalized, peak height, AMI, and BMI are calculated. An empirical p_value for the AMI is calculated by repeated random permutation of features. Graphical and quantitative results for each genomic feature are placed in a user specific session cache and returned to the user via AJAX as analysis of each feature completes.