d image, so for each micrograph, 16 correlations 
were assessed. For each prototype, we investigated the effect of varying the ST. By analyzing the 6 representative images of the training set with a total of 282 SGs, we empirically determined that a ST of 0.89 for prototype I; 0.88 for prototype II; 0.8 for prototype III; 0.88 for prototype IV; 0.86 for prototype V; 0.86 for prototype VI; 0.92 for prototype VII; and 0.85 for prototype VIII; eliminated all false positives in each case. As expected, these strict ST values reduced the number of positive hits. This was compensated by the use of several prototypes. We found that with the above indicated ST, each prototype identified a fraction between 2% and 32% of the SGs present in the images. We found that with the 8 proposed prototypes the script recognized all the SGs present in the original micrographs of the training set. However, as expected, a single SG was frequently recognized by more than one prototype. However, we found that object size was better measured when it was above 6 pixels in width, thus higher resolution and/or higher magnification images are required for accurate measurements of granule size. However, the object size in the micrograph is largely affected by the Point Spread Function of the objective and this may represent a serious limitation to size measurements by confocal microscopy. In summary, BUHO informs the cell number, the proportion of cells with SGs, the number of SGs per cell and their size with minimal deviation from manual-counted values. In addition, the algorithm calculates the distances between equal or distinct objects stained simultaneously using the bwdist function. exposed to thapsigargin, a known inductor of ER-stress, as indicated in Materials and Methods. To challenge our automated method with different data input, SGs were visualized by immunofluorescence against eIF3g, an accepted SG marker. Images were taken at 636, 102461024 pixels and scaled down to 512 pixels previous to analysis with the prototypes IVIII as above. As before, BUHO allowed us to assess transient SG formation with high precision. Deviation from manuallycounted values was less than 7% and both analysis showed a maximal response at 1 hour, when 93% of the cells had SGs. We concluded that BUHO perform successfully in the identification of SGs in cells stained by FISH against polyadenylated RNA or immunofluorescence against eIF3g. Moreover, the same 8 prototypes are suitable for SG identification in mammalian or Drosophila cells exposed to different insults. Example applications To assess the suitability of BUHO for the analysis of additional cell structures, we investigated several examples of cytoplasmic foci stained with different strategies, including SGs and PBs in mammalian and insect cells, and synapses and Smaug 1-mRNA silencing foci in primary neurons. Finally, we used BUHO for the analysis of a pilot experiment aimed to identify new genes regulators of SG formation. In all cases, we compared the BUHO output with values Seliciclib price obtained manually and found that differences were lower than 5% of the values. Example II: Analysis of PB induction upon stress PB size increases upon cellular stress, and we used this as a model system to test the BUHO suitability to assess changes in foci size. Drosophila S2R+ or mammalian U2OS cells were exposed to oxidative or ER-stress as indicated in Materials and Methods, and PBs were visualized by immunostaining of the specific marker DCP1a. Dro
