Figure 2

iCluster workflow to test for difference in imaging. (1) Treated/untreated sets of images are captured using identical microscope settings. (2) Images are stored in a directory together with a simple text format file describing which experiment each image belongs to. (3) The image description file together with the images is loaded into iCluster via a file selector. If (optional) image statistics have not been supplied by the user in the description file, TAS are generated automatically. Initially images are randomly placed in 3D. (4) The user initiates spatial layout by statistical similarity by clicking on 'Sammon Map'. (5) Once layout has finished, outlier images are found, viewed in detail, reclassified as 'Outliers', and removed from view. (6) Representative images for each experiment are automatically found and viewed in detail. (7) p-values are then calculated for the null hypothesis: no shift in localisation has occurred. For experiments where a large number of treatments are imaged, an initial test run on one or more treatments might be used to determine a minimum number of images required to detect difference. See also Supplementary Movie.