Ence of a carcinoma and microarray analysis was performed. Microarray data

Ence of a carcinoma and microarray analysis was performed. Microarray data used in the current analysis were from GEO (GSE21264). 31 mice were analyzed; for each of these mice we have data for all 3 progression steps (normal, papilloma, and carcinoma). The mouse ID numbers were the same IDs as in Quigley et al. [17] Genes were selected for analysis based 25033180 on detection and fold change. The starting data set represented 45,101 probe sets. The expression value of every gene in papilloma (P) and carcinoma (C) was normalized to the average expression value of the same gene in 80-49-9 normal (N), and the resulting ratios were transformed to log2. For the average analysis, T-test was then invoked, comparing each of P and C cells to N, and only genes with p value,0.05 were analyzed further (significant genes). Genes showing greater than 4-fold change were assigned to functional groups using DAVID software and KEGG database. Overrepresented gene categories were identifiedusing DAVID software. Results were filtered to remove categories with EASE score (a conservative variant of the one-tailed Fisher’s exact Peptide M probability) more than 0.001 and FDR.0. Pathways showing greater than 4-fold change according to DAVID were detected by the KEGG database. Data were analyzed in 2 ways: 1. “Average analysis”: For each transcript on the microarray, the expression over all 31 mice at each stage (normal, papilloma and carcinoma) was averaged. Ratios of expression between the stages, i.e. papilloma/normal (P/N), carcinoma/papilloma (C/P), carcinoma/normal (C/N) were calculated and filtered to select transcripts that showed at least a 4-fold change and had a Pvalue#0.05 in a T-test, to get a list of transcripts with significant changes. 2. “Heterogeneity analysis”: To compare the changes in gene expression during cancer progression between different individuals, we analyzed each mouse separately, using the expression data for normal skin, papilloma and carcinoma from that specific mouse. For each transcript, P/N, C/P and C/N were calculated and those transcripts that had at least a 4fold change were chosen for further analysis.Figure 1. Analysis of individual mice reveals diversity in transcript number that changed during carcinogenesis. The number of transcripts that were up-regulated (left panel) or down-regulated (right panel) at least 4-fold in each mouse during the transition from: A. normal skin to carcinoma (C/N); B. papilloma to carcinoma (C/P); and C. normal skin to papilloma (P/N). Mouse IDs refer to the IDs in the original data [17]. doi:10.1371/journal.pone.0057748.gHeterogeneous Gene Expression in SCC DevelopmentFigure 2. Heterogeneity in DAVID analysis of individual mice. The Y axis represents the number of mice in the group, and the X axis represents the number of significant annotations that were increased (left panel) or decreased (right panel) in each group. The annotations were examined in: (A) carcinomas vs. normal skin (C/N); (B) carcinomas vs. papillomas (C/P), and (C) papillomas vs. normal skin (P/N). Only 4 annotations were increased in C/N in all 31 mice, while 126 annotations were each increased in only one mouse. doi:10.1371/journal.pone.0057748.gDAVID analysisThe lists of genes with 4-fold change were inserted into DAVID for annotation analysis. Only DAVID annotations that had Pvalues equal to or less than 0.001 were considered further. The annotations lists for the individual mice were analyzed using an algorithm that enabled us to examine w.Ence of a carcinoma and microarray analysis was performed. Microarray data used in the current analysis were from GEO (GSE21264). 31 mice were analyzed; for each of these mice we have data for all 3 progression steps (normal, papilloma, and carcinoma). The mouse ID numbers were the same IDs as in Quigley et al. [17] Genes were selected for analysis based 25033180 on detection and fold change. The starting data set represented 45,101 probe sets. The expression value of every gene in papilloma (P) and carcinoma (C) was normalized to the average expression value of the same gene in normal (N), and the resulting ratios were transformed to log2. For the average analysis, T-test was then invoked, comparing each of P and C cells to N, and only genes with p value,0.05 were analyzed further (significant genes). Genes showing greater than 4-fold change were assigned to functional groups using DAVID software and KEGG database. Overrepresented gene categories were identifiedusing DAVID software. Results were filtered to remove categories with EASE score (a conservative variant of the one-tailed Fisher’s exact probability) more than 0.001 and FDR.0. Pathways showing greater than 4-fold change according to DAVID were detected by the KEGG database. Data were analyzed in 2 ways: 1. “Average analysis”: For each transcript on the microarray, the expression over all 31 mice at each stage (normal, papilloma and carcinoma) was averaged. Ratios of expression between the stages, i.e. papilloma/normal (P/N), carcinoma/papilloma (C/P), carcinoma/normal (C/N) were calculated and filtered to select transcripts that showed at least a 4-fold change and had a Pvalue#0.05 in a T-test, to get a list of transcripts with significant changes. 2. “Heterogeneity analysis”: To compare the changes in gene expression during cancer progression between different individuals, we analyzed each mouse separately, using the expression data for normal skin, papilloma and carcinoma from that specific mouse. For each transcript, P/N, C/P and C/N were calculated and those transcripts that had at least a 4fold change were chosen for further analysis.Figure 1. Analysis of individual mice reveals diversity in transcript number that changed during carcinogenesis. The number of transcripts that were up-regulated (left panel) or down-regulated (right panel) at least 4-fold in each mouse during the transition from: A. normal skin to carcinoma (C/N); B. papilloma to carcinoma (C/P); and C. normal skin to papilloma (P/N). Mouse IDs refer to the IDs in the original data [17]. doi:10.1371/journal.pone.0057748.gHeterogeneous Gene Expression in SCC DevelopmentFigure 2. Heterogeneity in DAVID analysis of individual mice. The Y axis represents the number of mice in the group, and the X axis represents the number of significant annotations that were increased (left panel) or decreased (right panel) in each group. The annotations were examined in: (A) carcinomas vs. normal skin (C/N); (B) carcinomas vs. papillomas (C/P), and (C) papillomas vs. normal skin (P/N). Only 4 annotations were increased in C/N in all 31 mice, while 126 annotations were each increased in only one mouse. doi:10.1371/journal.pone.0057748.gDAVID analysisThe lists of genes with 4-fold change were inserted into DAVID for annotation analysis. Only DAVID annotations that had Pvalues equal to or less than 0.001 were considered further. The annotations lists for the individual mice were analyzed using an algorithm that enabled us to examine w.