2.3. Data Integration by Meta-Analysis

Datasets selected for integration had a similar experimental design, sample size, and chemistry. These datasets were then merged using a modified Fisher’s combined p-value meta-analysis implemented through MetaDE R package [12], as was described by Rhodes et al. [13]. For each gene in every dataset, a p-value was determined by a t-test, after a p value modified (P-mod) was calculated by multiplying the −log10 (p-value) times log1.5 (absolute fold change). Xiao et al. [14] described in detail the p-value modification using this methodology. This modification allows the p-value to be enriched with the FC magnitude and provides better control of false positives. The P-mods of each gene in all datasets were combined using the Rhodes methodology.