Identifying direct contacts between protein complex subunits from their conditional dependence in proteomics datasets
Fig 3
The direct contact prediction method makes high confident predictions for well-sampled complexes.
A. Distribution of protein subunit sampling for complexes with a max direct contact probability > = 0.5 (blue) and complexes with a max probability < 0.5 (green). Sampling is measured for a complex by averaging the number of fractions for which each protein in the complex is observed (i.e. nonzero fractions). Our method performs better for well-sampled complexes and is limited for poorly sampled ones. B. Distribution of the mean number of nonzero fractions for pairs of proteins predicted by the direct contact method. Pairs of proteins with high probabilities are well sampled compared to those with lower probabilities.