This dataset was further processed first by removing compounds classified to more than a single target, then by removing targets with only one or two compounds and finally by subjecting it to data curation.
This dataset was further complemented by the results of automatically querying the Twitter search API every 10 minutes for tweets containing one or more of the keywords we identified as related to American Idol.
This dataset was further reduced to 278 sites, eliminating 17 because geographical coordinates were not available for these sites when the data were analyzed.
This dataset was further processed to remove redundancy, as in several instances, the precursor sequences are almost identical except for a few changes in nucleotides.
This dataset was further complemented with cysteine synthase M (cysM) protein sequences from bacteria and cysteine synthase protein sequences from fungi, Chromalveolata, plants, nematodes, and Planococcus citri and its three best BLASTp hits, harboring a cysteine synthase CDD motif (COG0031 or PLN2565).
This dataset was further reduced to those showing a >2-fold deviation from the mean of all five conditions in at least one sample, and by excluding all genes encoding VSG/VRs and ESAG/GRESAGs (which were analyzed separately, see below).
During the BioCreative IV evaluation period, the dataset was further divided into three subsets, which were the training set (3,500 abstracts), the development set (3,500 abstracts) and the test set (3,000 abstracts).
The signal intensities in the dataset were further log transformed (base 2), and missing values were replaced by the average of nearest rows neighbour values.
Since the Anova approach is not ideal for the analysis of several simultaneous measurements, the dataset was further analyzed using multivariate analysis.
