Because we were interested in carryover effects of a voice on the consecutive one, we computed the reaction times as a function of physical difference between 2 consecutive stimuli (Fig. 1 d ).
Since c ∗21.6 > 1, we compute the one-sided confidence interval with critical value 4.45.
Using all the 4,291 complete PrSMs with one PTM, we computed "estimated" FDRs for cut-off p-values in {0.0001, 0.0002,..., 1.0000} based on spectral probabilities.
To quantify the location shift between any two laps, for each sequence in one lap we computed the distance between its center location and the center location of the nearest sequence in the other lap.
Finally, as the vertices in the approximate ancestry graph G correspond to strongly connected components that typically include more than one mutation, we compute the frequency matrix F = [ f p j ] for the approximate ancestry graph G by combining read counts for all mutations in the same component.
For both 5-fold validation on pooled data and leave-one-stent out, we computed detection statistics listed below.
That's the one for which we computed the energy, but there's something wrong with our argument.
For one-back performance, we computed the change (ΔET) in the frequency of target selection errors (ET) from the OB condition to the OB+ROT condition (ΔET = ET OB+ROT) − ET OB)).
For samples with more than one polymorphic amplicon we computed the average of the separate CNs or of their least common multiple.
Based on TCGA Infinium 450K data, for any DMR identified in our study that overlaped at least one Infinium probe, we computed the average methylation level (aML) in each cancer type group (described in Additional file 1: Table S8) using the beta value of overlapping Infinium probes.
