Statistical and visual differentiation of subcellular imaging
1 Institute for Molecular Bioscience, The University of Queensland, Brisbane, Australia
2 ARC Centre of Excellence in Bioinformatics, The University of Queensland, Brisbane, Australia
BMC Bioinformatics 2009, 10:94 doi:10.1186/1471-2105-10-94Published: 22 March 2009
An example of using iCluster. Initially, 50 mitochondria images (mitotracker) and 50 plasma membrane images (EGFR) are shown randomly placed having been loaded into iCluster and statistics calculated. 'Sammon Map Statistics' is then selected and the images move around as a spatial layout is found that reflects the distances between the statistics vectors for the images. The user then rotates the image set, and 3 outlier images are observed, selected (red tint), and then show in more detail in a 2D representation. All three appear to contain artefacts. The view then switches back to the 3D view, a new class 'outlier' is added to the class list, the selected images are reclassified to this class (green borders), and then removed from view by deselecting their class button. Representatives for each class of the remaining images are then shown side by side in a 2D view. The view then changes back to the 3D view, and 'Statistical Test' selected. The images to compare and the number of repeats to calculate a p-value for the null hypothesis (no difference) are then selected. Finally, the returned p-value of 0.000 is displayed, showing that the visual assessment of difference is confirmed statistically.
Format: MOV Size: 3.7MB Download file