An Improved, Bias-Reduced Probabilistic Functional Gene Network of Baker's Yeast, Saccharomyces cerevisiae
Each plot presents the correlation (for a given network) between network centrality, calculated as the number of interactions per gene normalized by the maximum observed value, versus the essentiality of the genes, calculated as the fraction of essential yeast genes  for each successive bin (open circle) of 100 genes ranked by decreasing degree centrality. (A) shows the trend for a high quality protein-protein physical interaction network derived from DIP  and bioGRID , (B) shows the trend for YeastNet version 1 (34,000 most confident linkages only), and (C) shows the trend for YeastNet version 2. For both functional networks, degree centrality is weighted by the interaction LLS scores (i.e., calculated as the sum of LLS scores for a gene, divided by the maximum sum of LLS scores observed in the network). The degree of correlation is measured as the Spearman rank correlation coefficient (rs).