Supplementary Figure 8: The Functional Representation module is robust to data compendium size, amount of prior knowledge, and initialization state.
From: An integrative tissue-network approach to identify and test human disease genes
Semi-supervised network construction approach was applied to (a) ten progressively smaller compendia sub-sampled from the full worm compendium (without replacement) and (b) seven progressively smaller sets of tissue gene annotations subsampled from all previously known tissue genes (without replacement). Each measurement is an average of 10 independent simulations and standard error (shaded regions) is shown.
