Fig. 5: Functional links between catalytic enzymes and significantly often co-depleted proteins reveal differences in protein palmitoylation status in aggressive LUAD subtype.
a Left: Fraction of depleted proteins annotated as either 1st (sky-blue) or 2nd degree (blue) SH interactors. Right: Frequency of depleted proteins found in 200 random interactomes. Red arrow indicates the number of depleted proteins in the real SH interactome (N = 258), which significantly exceeds the number of binary SH interactors compared to random sets. One-tailed p-value calculated by pnorm-test. b Left: Line charts show activity and depletion of respective “enzyme-protein” pairs that correlated between samples and demonstrated physical PPI evidence. Two-sided p < 0.05 from Correlation coefficient significance test (n = 12 paired samples). Green and blue lines depict either average SH activity or average protein depletion across samples, respectively. The x-axis correspond to adjacent-tissue (N) or tumor (T) samples. Right: Histogram depicts the number of physical interactions for 100 random subsets generated by sampling of quantified proteins. Red arrow is the number of confirmed physical interactions found within subsets of co-depleted proteins and SHs displaying significant pairwise correlations (N = 55). One-tailed p-value from pnorm-test in R. c Left panel: Sample classification based on consensus model with 75% confidence ellipses and PLS-DA variates 1, 2 and 3. Consensus model obtained by block PLS-DA data integration and successive tuning steps (i.e., 82 classifiers), integrating 23 respective SHs and 59 significantly often co-depleted proteins (Supplementary data Fig. 8c). d Functional enrichment analysis of classifiers from consensus model in PANTHER Overrepresentation Fisher Exact Test. FDR adjusted p-value for multiple testing. Homo sapiens database is used as reference. e Left panel: Unsupervised sample clustering based on final consensus model. Manhattan distances used for hierarchical clustering. Right panel: Network depicts 9 hydrolases (i.e., SH in green) and subset of co-depleted proteins (i.e., DP in blue) including those with observed S-palmitoylation motif (asterisk in orange). The correlation between data levels extracted after performing block PLS-DA (Sparse Generalized Canonical Correlation Analysis). Blue hexagon and green triangle depict co-depleted proteins either with mitochondrial localization or involved in FA metabolism, respectively. Source data are provided as a Source Data file.
