13 Both studies contrasted samples from donor lungs that later developed PGD against donor lungs that did not. For the GSE9102 study, cDNA microarray data as pre-processed by the authors were used, and covered expression measurements for 6727 Ensembl build 55

human genes (http://jul2009.archive.ensembl.org). When several probes were available for the same gene, the probe displaying the most significant differential XAV-939 expression was selected to represent that gene. For the GSE8021 study, the original raw data were processed as follows. Affymetrix Human Genome U133A 2.0 Array probes were remapped to 11894 different Ensembl build 55 human genes.14 Using these redefined probe sets, probe intensities were summarized and made comparable between arrays by quantile normalization as implemented in the Robust Multi-Array Average expression measure.15 It was possible to identify corresponding gene expression for 242 of the 272 proteins on the antigen microarray (89%). For each antigen and detection antibody, differential reactivity between patients without PGD (n = 19) and patients with PGD (n = 20) was evaluated by calculating ratios (fold-changes), t-statistics and P-values. For each gene measured, differential expression between donor lungs developing PGD (16 and 10) and those that did not (34 and 16) were similarly evaluated by

ratios, t-statistics and P-values. Multiple testing was controlled using the false discovery rate.16 A human protein interaction network was created by pooling human interaction find more data from several of the largest databases.17 Coverage was further TCL increased by transferring data from model organisms. A network-wide confidence score for all interactions, based on network topology, experimental type and interaction reproducibility, was then established. The reliability of this score as a measure of interaction confidence was confirmed by fitting a calibration curve of the score against a high-confidence set of about 35 000 human interactions. As previously described,8 all interactions with a confidence score above 0·154

were included, resulting in a network containing approximately 154 000 unique interactions between approximately 12 500 human proteins. Out of the 272 proteins on the antigen microarray, 260 (96%) were among these. As described previously,8 the statistical significance of the number of proteins in a network (the size) extracted from a given larger set of proteins, was estimated by randomly selecting sets of proteins of the same size, each time recording the size of the largest network possible to extract. For 107 such randomizations, the proportion of random sets of proteins for which equally sized or larger networks could be extracted, establishes the P-value of the network extracted from the original protein set. Over-represented biological processes among proteins in networks were identified by hypergeometric testing of gene ontology terms.