13 This yielded a total of 13,016 peptides with adequate peptide intensity reproducibility. Technical replicates were aligned via regression on the log10 intensity measurements and averaged to sample level estimates. The dataset was then normalized using a local mean centering based on a rank invariant peptide subset of 185 peptides, and the resultant data are presented in Supporting Table 3. Patients were grouped into the liver disease progressor or nonprogressor category based on the presence or absence of histologic evidence of

severe liver injury (Batts-Ludwig fibrosis stages 3-4) at 1 year posttransplantation (Fig. 1). Identification of significantly differentially expressed proteins among patient groups was performed by computing area under the receiver operating characteristic (ROC) find more curve for binary comparisons, and visualized using singular value decomposition initialized multidimensional scaling (SVD-MDS)14, 15 as described in the Supporting Materials and Methods. Samples were collected in 10-mL vacutainers (Becton Dickinson) and allowed to sit at room temperature for 15-30 minutes to allow clotting. The samples were then centrifuged at 3,000g for 15 minutes, the resultant

supernatant aliquoted into CryoTube Vials and stored at −80°C until use. Unbiased metabolomic profiling using 100 μL of serum was performed by Metabolon (Metabolon Inc, Durham, NC) using liquid/gas chromatography coupled with mass PF-02341066 ic50 spectrometry as described.16-18 Coabundance networks that relate proteins by similarity in their abundance profiles (patterns of expression across all patients) allow representation of system-level patterns in the data. A protein association network was constructed based on correlations between protein abundance such that proteins exhibiting similar abundances are connected in the network.12,19 We then integrated information on known protein-protein interactions (http://cytoscape.wodaklab.org/wiki/Data_Sets), producing a master network of connected learn more cellular processes. The topology (connectivity of proteins) in the network

was then calculated using the igraph library in R. The connectivity of proteins or genes in biological networks can provide insight into their relative importance. Briefly, protein or gene “hubs” exhibiting a high degree of connectivity (connected to many other proteins or genes) and “bottlenecks” exhibiting a high betweeness (key connectors of subnetworks within a network) represent central points for controlling communication within a network and tend to play an essential role in growth, virulence and targeting by pathogens.12, 19-22 Bottlenecks were considered to be the top 20% of proteins as ranked by betweeness,20-22 though all observations were similar using 10% and 5% thresholds. Statistical significance was calculated using a Fisher’s exact test; P < 0.05 was considered significant.