Fig. 4From: Determining the optimal number of independent components for reproducible transcriptomic data analysisConservation or non-conservation of independent components in higher than MSTD dimensions. ICA decomposition in MSTD data dimension is compared with the one of dimension 50 and 100. Table on the left: list of different scenario with the relative frequency of each of them estimated for a pan-cancer TCGA dataset. Plots on the right show the frequency of finding the reciprocally correlated component in the higher dimension and the average correlation coefficient of the reciprocal correlation, as a function of the relative rank of the component (component rank minus MSTD value for a given dataset)Back to article page