For comparison with gene expression measured in whole-tissue samples, we considered the set of probes that were present on the two array chips.
To identify minimal-size genesets capable of optimally discriminating between malignant and normal samples, we considered the expression values for the set of 318 genes in these 84 samples.
For the lower surface of the sample, we constrain all displacements, u x = u y = u z = 0. Regarding the free upper surface of the sample, we consider the electrochemical potential of water and ions identical to those measured in the external bath, ε w = ε w *; ε+ = ε+*; ε− = ε−*.
While our strand bias test could not improve ShoRAH's raw SNV calls for the HIV sample, we consider the slight reduction in recall with σ = 0.0111 (two additional very low frequency true SNVs were classified as errors) to be an acceptable cost, compared to the potential benefits of the test.
In comparing the suitability of different submodels estimated by Gibbs sampling, we consider the popular deviance information criterion (DIC) (Spiegelhalter et al. 2002), defined as 2 × ℓ (E ^ ; y ) − 4 × E ^ (ℓ (θ ; y ) ), where θ collects all parameters, ℓ(·| y) is the log-likelihood, and E ^ is the mean over all T iterations.
For double FIT sampling, we considered the FIT positive when the cut-off level exceeded 50 ng/ml in at least one out of two samples.
In view of the low estimates of heritability found in this sample, we considered the sample size to be too small to attempt a partitioning into training and validation datasets.
To calculate the probability sample, we considered the situation of higher probability of sampling error (p =0.50) assuming a confidence level of 95% and a sampling error of 5%.
In generating a set of samples, we consider only the variations of parameter values in the conditional distributions (Tables 2 (right group) and 3 (upper right).
Additional comments need to be made regarding the samples we considered in the analysis.
