Skip to main content
Fig. 6 | Biology Direct

Fig. 6

From: Quantitative proteomics signature profiling based on network contextualization

Fig. 6

Stability analysis of qPSP and hyp_geo using bootstrap resampling in CR (Colorectal Cancer). a Distribution of number of significant complexes returned. Across various sampling sizes (4, 6 and 8), qPSP consistently reported more significant complexes than t-test selection for differential proteins followed by complex selection using the hypergeometric test (hyp_geo). The difference was sufficiently large to obviate the need for a statistical test. b Simulation similarity comparisons. Pair-wise analysis of simulations to calculate the agreement levels (using Jaccard Score, 0 for complete disagreement, 1 for complete agreement) across complexes showed that qPSP was far more consistent than hyp_geo. c Complex persistency distribution. Distributions of significant complex agreements (On the x-axis, a score of 1 means complete persistence across all simulations, the y-axis is a frequency measurement, and its sum adds up to all complexes observed to be significant at least once). As sampling size increased, more complexes became more persistently represented across all simulations. However, this effect was much more pronounced for qPSP than hyp_geo. More importantly, the left skew for qPSP showed that most complexes were stable across samplings, while the right skew for hyp_geo showed that most of the significant complexes picked up were unstable

Back to article page