Additional file 1: of An ensemble framework for identifying essential proteins

Figure S1. The distributions of node strength for essential and nonessential protein. Figure S2. The distributions of co-expression weights for IBEPs and Non-IBEPs. Figure S3. Performance comparison of five centrality measures (BC, CC, DC, EC, and SC) on two yeast PINs (PIN24K and PIN76K) using uniform thresholding strategy ((a)-(h)). Figure S4. Relationship between the number of nonzero-degree nodes (or proteins) in PINs and the thresholds for generating the corresponding PINs using absolute thresholding strategy ((a)-(b)). Figure S5. Relationship between the number of nonzero-degree nodes (or proteins) in PINs and the thresholds for generating the PINs using uniform thresholding strategy ((a)-(b)). Figure S6. Performance comparison of five centrality measures (BC, CC, DC, EC, and SC) with their corresponding ensemble methods (absolute thresholding strategy) with different sample sizes or weights on two yeast PINs. Figure S7. Comparison of the number of essential proteins detected by each ensemble method using uniform thresholding strategy with different voting weights on two yeast PINs. Table S1. The number of common predicted proteins (overlap) among the top 100 proteins ranked by PCC-weighted methods. Table S2. The number of common predicted proteins (overlap) among the top 100 proteins ranked by single PCC-threshold methods (thr = 0.75). Table S3. Correlation between centrality measures based on their top 100 ranked proteins. Table S4. Correlation between ensemble methods based on their top 100 ranked proteins. (PDF 1294 kb)