Dataset: Transcription profiling of human bladder cancers to develop a clinical classification according to microarray expression profiles
Using Affymetrix microarray technology we analyzed the gene expression profiles of the most important pathological categories of bladder...
Using Affymetrix microarray technology we analyzed the gene expression profiles of the most important pathological categories of bladder cancer in order to detect potential marker genes. Applying an unsupervised cluster algorithm we observed clear differences between tumor and control samples, as well as between superficial and muscle invasive tumors. According to cluster results, the T1 high grade tumor type presented a global genetic profile which could not be distinguished from invasive cases. We described a new measure to classify differentially expressed genes and we compared it against the B-rank statistic as a standard method. According to this new classification method, the biological functions overrepresented in top differentially expressed genes when comparing tumor versus control samples were associated with growth, differentiation, immune system response, communication, cellular matrix and enzyme regulation. Comparing superficial versus invasive samples, the most important overrepresented biological category was growth and, specifically, DNA synthesis and mitotic cytoskeleton. On the other hand, some under expressed genes have been clearly related to muscular tissue contamination in control samples. Finally, we demonstrated that a pool strategy could be a good option to detect the best differentially expressed genes between two compared conditions. Experiment Overall Design: We analyzed gene expression profiles in normal bladder tissues (controls), low grade superficial tumor samples (pathologically classified as Ta low grade, named as Ta), high grade superficial tumors with an unclear clinical behavior (T1 high grade, named as T1) and high grade muscle invasive tumors (pathologically classified as T2, T3 or T4, named as T2+). We analyzed data using a sub-pooling strategy. The number of individual samples on every pool was: controls (4, 4, 4), Ta (5, 5, 5), T1 (5, 4, 4) and T2+ (5, 5, 5).
- Dec.12, 2014
- Sep.22, 2014