br Gene set analysis br To
Gene set analysis
To quantify the extent to which different groups of genes were enriched in a given metric (e.g., p values from a DE analysis), a gene set analysis (GSA) was performed. This type of analysis was applied in a number of situations throughout the study, and followed the same procedure (described below), unless stated otherwise. The following gene set collections were retrieved from the Molecular Signatures Database (MSigBD (Subramanian et al., 2005)): hallmark (H) (Liberzon et al., 2015), KEGG (C2 CP:KEGG), Reactome (C2 CP:REACTOME), GO biological process (C5 BP), and GO molecular function (C5 MF).
Gene set collections were filtered to remove all non-secretome genes from each set prior to analysis, unless otherwise stated. We note that this Pyocyanin can cause the name of a gene set to become less representative if a substantial portion of genes in the set are removed. In this way, the significance of a gene set does not necessarily represent an enrichment in its named function/pathway, but instead represents an enrichment in the set of secretome genes that are associated with that function/pathway. In addition, gene sets containing more than 400 genes (before filtering) were also removed, as these sets tended to have a very low fraction of secretome genes, and were generally uninformative. Finally, to avoid statistical problems with very small gene sets, those with less than 20 genes after filtration were excluded from the analysis, unless otherwise noted.
A Wilcoxon rank-sum test statistic was calculated from the DE analysis p values of genes in a given set, and compared to those of 100,000 randomly shuffled gene sets of the same size. The significance (p value) of a gene set was calculated as: p = 1 + NrandRset (3)
where Nrand R set is the number of randomly shuffled gene sets with a test statistic greater than or equal to that of the original gene set, and Nperms is the number of random permutations (100,000 in this study). Gene set p values calculated in this manner correspond to ‘‘non-directional’’ p values (pnon-dir), as they do not take into account the direction (increase or decrease) of the fold-change from the DE analysis, only the significance.
‘‘Distinct directional’’ gene set p values (pdist-dir-up and pdist-dir-down) (Va¨remo et al., 2013) were obtained in the same manner, except the p values from the DE analysis were first converted to directional p values (Equation 2) before calculating the associated Wilcoxon test statistic. The resulting gene set pdist-dir-up values quantify coordinated expression increases in a gene set, where a low pdist-dir-up
indicates a get set that is enriched in genes with significant expression increases. Coordinated expression decreases are quantified simply as pdist-dir-down = 1 – pdist-dir-up, where low pdist-dir-down values indicate an enrichment of genes with expression decreases.
Adjustment of p values All adjusted p values (padj) reported in the study were adjusted to control for the false discovery rate (FDR) using the Benjamini-Hoch-berg procedure. Statistical significance in this study was defined as padj < 0.05.
A systematic review and meta-analysis of stereotactic body radiation therapy versus surgery for patients with non–small cell lung cancer
Christopher Cao, MD, PhD,a,b Daniel Wang, MD,c Caroline Chung, MD,c David Tian, MD,b Andreas Rimner, MD,d James Huang, MD,a and David R. Jones, MDa
Objective: Stereotactic body radiation therapy is the preferred treatment modality for patients with inoperable early-stage non–small cell lung cancer. However, comparative outcomes between stereotactic body radiation therapy and surgery for high-risk patients remain controversial. The primary aim of the present meta-analysis was to assess overall survival in matched and unmatched patient cohorts undergoing stereotactic body radiation therapy or surgery. Secondary end points included cancer-specific survival, disease-free survival, disease recurrence, and perioperative outcomes.