This procedure was repeated for each sample, computing the statistical significance of each gene in the various leave-one-out datasets.
The protector aligns these L samples with the first L received samples, computes the correlation, shifts the alignment by one sample and then re-computes the correlation.
The second method (MHS: median of HapMap samples) computes the reference using the median of external control samples (in this case they are from HapMap).
In the cases that we studied, this procedure increased the computational time of each sample by a factor of 5 when compared with a sample computed using the standard Monte Carlo.
For every bin, the per-million reads coverage of each sample was normalized using the size factor of the sample computed by DESeq [ 57], and replicates were averaged.
For these three sample sets (19,000 DE samples, 19,000 LH samples and 100 LH samples) we computed the acceptance ratios of all 119 experimental strains and sorted them in ascending order, as shown in Figure 8.
We repeated this process for each sample and computed the averaged predictor error on the testing sample at each grid point λ ∈ {10−3, 10−2, 10−1, 1, 10}.
For each sample we computed the corresponding functional boxplot.
For each such permuted sample, we computed the mean, over all positive TC nuclei, of the distance to the closest inner cell nucleus.
To estimate the degree of discriminable variation in coloration within each plumage patch for our sample we computed the visual contrast (hereafter ΔSvar or ΔLvar) between each point and a fixed point in space.
