Identifying direct contacts between protein complex subunits from their conditional dependence in proteomics datasets
Fig 4
Relationship between direct contact probability and correlation for four example complexes with known structure.
A. Direct contact predictions for the well-observed proteasome (pdbid: 4CR2) show good discrimination between true (blue circle) and false positives (red x) compared to that of correlation. B. Direct contact predictions for the moderately observed spliceosome complex (pdbid: 5MQF) shows good discrimination between true and false positives but with a limited number of total predictions. C. The direct contact method does not make high confident predictions for the mitochondrial ribosome (pdbid: 4CE4) due to its limited sampling while correlation makes many high ranking false positive predictions. D. Similar to C, the direct contact method does not make predictions for the mitochondrial super-complex (pdbid: 2YBB) due to its limited sampling while correlation makes several high confident true positive predictions.