SUPPLEMENTARY MATERIAL

GENE EXPRESSION PATHWAYS OF HIGH GRADE LOCALIZED PROSTATE CANCER

Ashley E. Ross, Luigi Marchionni, Milena Vuica-Ross, Chris Cheadle, Jinshui Fan, David M. Berman, Edward M. Schaeffer

A compressed archive with all raw gene expression data can be downloaded here:

  • Download "raw" data

    The "Supplementary Methods" section can be downloaded here:

  • Download "Supplementary Methods" section

    LCM figures: large pdf file (~78Mb)

  • LCM data images

    Supplementary figure 1:
    Differentially expressed genes from laser captured Gleason grade 3 or 4 prostate cancer demonstrate increased concordance across platforms and more differentially expressed genes within platforms if pattern 4 disease is derived from the primary Gleason grade. Venn diagrams demonstrating the union of differentially expressed genes across arrays from Tomlins et al. (GSE6099) and True et al. (GSE5132). (A) Comparison of differentially expressed genes from Gleason pattern 3 prostate cancer laser captured form Gleason 3+3=6 cases versus Gleason pattern 4 prostate cancer captured from primary pattern 4. (B) Comparison of differentially expressed genes from Gleason pattern 3 and Gleason pattern 4 derived from primary, secondary or tertiary Gleason grade. The number of genes, concordantly induced in high grade prostate cancer is indicated in green and those concordantly repressed are listed in red. "Gene Space" refers to the total number of genes represented in the two platform comparison. P values correspond to the probability of discovering by chance that number of concordantly up-regulated (Pup, green) or down-regulated (Pdown, red) genes across platforms.

  • Supplementary figure 1

    Supplementary figure 2:
    Batch effects and non-concordance of Pressinotti et al. (GSE15484) with differentially expressed genes from data sets of True et al. (GSE5132) and Tomlins et al. (GSE6099). (A) Correspondence at the top plots for differentially expressed genes comparing prostate cancer to normal adjacent epithelium correlating data from Presinotti et al. with data from Tomlins et al. (blue) or True et al. (red), or comparing Tomlins et al. to True et al. (green). Lines below the 45 degree diagonal correspond to those having less agreement between differentially expressed genes than would be expected by chance. Lines in the white area of the graph have a probability of agreeing or disagreeing of P < 1E-06. Genes ranked by decreasing and increasing t-statistic are shown on the left and right respectively. (B) Pairwise correlation between samples labeled by date (y-axis) or sample group (benign, Gleason sum 6 (G6) or Gleason sum 8 or greater (G8), x-axis) demonstrating higher correlations among samples arrayed on the same date than of the same biological type. Two clear batches are evident.

  • Supplementary figure 2

  • Supplementary figure 2, panels A and B, high resolution

  • Supplementary figure 2, panel C, high resolution

    Supplementary figure 3:
    Venn diagrams demonstrating consensus differentially expressed between platforms. Number of concordantly differentially expressed genes (either up-regulated (green) or repressed (red)) comparing Gleason pattern 3 (from 3+3=6) and primary Gleason pattern 4 prostate cancer between array platforms (Ross et al. (this paper), True et al. (GSE5132) and Tomlins et al.(GSE6099)) are found within intersections. "Gene Space" refers to the total number of genes represented in the two or three platform comparison. P values correspond to the probability of discovering by chance that number of concordantly up-regulated (Pup) or down-regulated (Pdown) genes across platforms.

  • Supplementary figure 3

    Additional data:

    Links To our data Results Files Description
    1 Our data Gene Lists Gene lists of differentially expressed genes:
  • Comparison between pure G3 (Gleason score 6) and pure G4 (Gleason score 8)
  • Robust regression analysis using all genes and samples;
    • 2 Our data Figures Pre-processing and linear model figures
    3 Our data AFA results Results from Analysis of Functional Annotation for the linear model analysis


    Links To True et al 2006 (GSE5132) Results Files Description
    1 True (GSE5132) Gene Lists Gene lists of differentially expressed genes:
  • Comparison between pure G3 (Gleason score 6) and primary G4 (Gleason score 7, 8, and 9);
  • Comparison between Low Gleason score (pattern 3 from Gleason score 6 and 7) and High Gleason score (any pattern 4 from Gleason score 7, 8, and 9);
  • Comparison between Prostate cancer and Adjacent normal;
  • Least squares method using a single channel design and intraspot correlation: normal adjacent samples modeled as one group;
    • 2 True (GSE5132) Figures Pre-processing and linear model figures
    3 True (GSE5132) AFA results Results from Analysis of Functional Annotation


    Links To Tomlins et al 2007 (GSE6099) Results Files Description
    1 Tomlins (GSE6099) Gene Lists Gene lists of differentially expressed genes:
  • Comparison between pure G3 (Gleason score 6) and primary G4 (Gleason score 7, 8, and 9);
  • Comparison between Low Gleason score (pattern 3 from Gleason score 6 and 7) and High Gleason score (any pattern 4 from Gleason score 7, 8, and 9);
  • Comparison between Prostate cancer and Adjacent normal;
  • Comparison between Prostate cancer and Donor normal;
  • Comparison between Adjacent and Donor normal;
  • Robust regression was used with all genes and samples;
    • 2 Tomlins (GSE6099) Figures Pre-processing and linear model figures
    3 Tomlins (GSE6099) AFA results Results from Analysis of Functional Annotation


    Links To Pressinotti et al 2009 (GSE15484) Results Files Description
    1 Pressinotti (GSE15484) Gene Lists Gene lists of differentially expressed genes:
  • Comparison between Gleason score 6 and Gleason 8-10;
  • Comparison between Prostate cancer and Adjacent normal;
  • Robust regression was used with all genes and samples, with and without batch effect;
    • 2 Pressinotti (GSE15484) Figures Pre-processing and linear model figures


    Links To Meta-analysis Results Files Description
    3 Data sets meta-analysis Gene Lists, q-value < 0.2 Gene lists of differentially expressed genes across three data sets (q-value < 0.2):
  • Gene lists corresponding to Venn diagram intersections;
  • For our data: robust regression with ALL genes and samples;
  • For True et al 2006 (GSE5132): least squares method with single normal adjacent group;
  • For Tomlins et al 2007 (GSE6099): robust regression;
    • 3 Data sets meta-analysis Figures, q-value < 0.2 Data sets meta-analysis figures
    Cat-plots, Venn diagrams (q-value < 0.2), pairwise correlation plots)
    3 Data sets meta-analysis AFA results for the Venn diagram intersections Results from Analysis of Functional Annotation using cross-tabulation and the chis-squared test


    Page Counter

    Back to Publications Page
    Back to Home Page
    Back to Top