• Gene Discovery in Bladder Cancer Progression using cDNA Microarrays. Sanchez-Carbayo M, Socci ND, Lozano JJ, Li W, Charytonowicz E, Belbin TJ, Prystowsky MB, Ortiz AR, Childs G, Cordon-Cardo C. Am J Pathol. 2003 Aug;163(2):505-16.
  
  In this article gene expression changes through bladder cancer progression were identified using cDNA microarrays. The authors obtained a stage classification doing hierchichal clustering. Moreover, they recognised that early and invasive stages had similar genes’ profile. To identify relevant genes involved in bladder cancer progression, they used different statistical methods like for example t-test. They identified expresion patterns related on pathology and they also obtained relevant genes and signaling pathways involved in cancer progression. The main objective was the identification of molecular markers of bladder cancer.
  
  
  
• Identifying distinct classes of bladder carcinoma using microarrays. Dyrskjot L, Thykjaer T, Kruhoffer M, Jensen JL, Marcussen N, Hamilton-Dutoit S, Wolf H, Orntoft TF. Nat Genet. 2003 Jan;33(1):90-6. Epub 2002 Dec 09.
  
  In this article the authors identify different subtypes of bladder cancer using microarrays expression analysis. They classified cancer progression in three main stages: Ta, T1 and T2-4 using hierchichal clustering. They also obtained another classification based in the recurrence of tumours. Genes whose regulation is the key for superficial and invasive classification were identified. Characteristic expression gene patterns let them identify biological properties which are altered in human cancers.

Databases. We obtained the database of the article “Identifying distinct classes of bladder carcinoma using microarrays” from NCBI web page in Gene Expression Omnibus (GEO).

Bladder tumour stage classification is a simple channel database. We just wanted to study the progresion of cancer and this is why we only checked tumour samples Ta, T1 and T2+.

Significant expression values selection. In this step, we studied statistical analysis using R-script. We found this program in the following web page:nin.crg.es/cgi-bin/pMargeWeb.cgi.

First of all, we identified samples corresponding on different stages and we created three files: Ta-T1, T1-T2 and Ta-T2. We obtained p-values. P-value shows differential expression significance. In our study, we selected genes which had a p-value lower than 0.001 (p<0.001) following the next process:

Looking for “Gene Symbol”. We obtained gene symbols using EASE program. These symbols will be very important to analyze results lately.

EASE program was found in DAVID page (Database for Annotation, Visualization and Integrated Discovery). We introduced gene codes in “Input Genes” and we pressed “Annotate Genes”. Apart from obtaining gene symbols, we also obtained some information about functionality and chromosome location.

Clusterings.It is one of the most frequent method used to analyzing information obtained by biochips. This method displays the information in a tree cluster, which is very useful because we can see the results graphically. There is not only one type of clustering method. Here we show the classification and give information of different clusters:

• Hierchichal
  • Aglomerative
  • Hierchichal tree
  • Simple
  • Easy visualization Tree View
• K-means
  • Divisible
  • Number of genes predetermined
  • An expression mean vector is calculated
  • No trees are obtained
• SOMs (Self Organizing Maps)
  • Divisible
  • Based in neuronal connections
  • Optimisize groups separation using predefined geometry
  • Convergence between vectors of each cluster is maximized

In our project, we used hierchichal clustering. First of all, we went to GEPAS (Gene Expression Pattern Analysis Suite) and followed these instructions:

We chose Log-transform base 2 (changing negative values to 10 is important) and Standardize patterns. This process normalize our values, which is very important to get good results.

Afterwards, we send to cluster. We selected UPGMA using correlation distance condition.

Genes functionality. To find information about the function of genes and biological processes in which these genes were involved, we used two sources::

• DAVID. In this web page, we went to GOCharts. This program is very useful because it shows graphically gene’s function. We decided to check biological processes and mollecular function because it was the information more interesting for us. We chose level 5 of specificity and 3 for minimal hits (except T1-T2 where we only selected 1 minimal hit because there were just a few genes differentially expressed from T1 to T2).
• FatiGO. We defined the organism: human, we chose mollecular function and we wrote our list of genes. From FatiGO, we obtained graphic results of representative functions.

Representative mRNA Access. We wanted to have the same kind of gene nomenclature to be able to compare and analyze the results. For this, we used SOURCE program.

Compare results.

In the previous step, we obtained files containig two columns, one of them had gene symbols and the other one contained corresponding Representative mRNAcess for each gene. To compare our results and the article’s ones, we needed files with just one of the columns. That was the reason why we used these commands in Shell (Unix): Then, we ordered every list of genes using sort and we eliminated repeated genes using uniq command to avoid mistakes:

Finally, we should identify genes that were common in both studies. We determined if we had obtained the same representative genes than Sanchez-Carbayo or if there were no common genes.

Now, we are going to comment the results that we have obtained:

Gene Discovery in Bladder Cancer Progression using cDNA Microarrays.
According to Sanchez-Carbayo article, significant overexpressed genes would be related on:

• Excretion
• RNA Splicing
• Neurogenesis
• Development
• Attachment proteins
• Lipid metabolism (genes maybe involved in angyogenesis)
• Differentiation epidermical signals like keratin 15
• Immunological response
• Extracellular matrix remodelling
• Oncogenesis

• Cellular adhesion
• Protein modification
• Development+
• Immunological response
• Lipid metabolism
• Attachment proteins
• Oncogenesis
• Transcription regulation
• Cell cycle (p53, pRb, p21)

There are other pathways with a differential expression patterns. These are: checkpoint regulation (for example Cdc16, which is necessary in mitotic spindle formation), and also genes involved in apoptosis.

Identifying distinct classes of bladder carcinoma using microarrays.

Comparing gene expression in different stages they conclude that:

Overexpressed genes are related on:
• transcription
• Nuclear processes
• Proliferation
• Matrix remodelling
• Repressed genes:
  • Codify extracellular matrix proteins.
  • Are involved in immunologic system

Our results.

Here we show clusters that we have obtained by taking genes with p-value lower than 0.001:

Cluster progressió Ta-T1

Cluster progressió T1-T2

Cluster progressió Ta-T2

*You can see images bigger if you click on them.

As we said before in Matherial and Methods, using Shell commands we tried to find genes which were differentially expressed in Sanchez-Carbayo article and in the database of the other article that we analyzed.

Unfortunately, we did not find common genes. Then, we focused our objectives on analyzing the genes which were more significative according to clusters. Below we show chosen genes, their localisation and also their functions.

Genes differentially expressed in Ta-T1 progression:

Genes differentially expressed trogh Ta-T2 progression:

As we can verify, significant genes wih p-values<0.001 have similar functions as compared with representative genes of both articles. Therefore, we can see genes typically related on tumoral processes as for instance genes involved in cell adhesion and genes responsible for decreasing immune system efficacy. We can also see changes in lipid metabolism and deregulation of transcription driving to cell cycle regulation problems.