The document summarizes a thesis that aimed to identify gene expression modules in colorectal cancer using three different methods. The results conclusively identified functional gene expression modules and mapped them to known pathways. Some modules predicted tumor relapse in colorectal cancer patients and survival in breast cancer patients. The conclusions state that gene expression modules regularly occurring in colorectal cancer were identified and their functional significance was found.

1. Identification of Gene Expression Modules in Colorectal Cancer September 08, 2009 Manikhandan A V Mudaliar MSc Thesis Supervisors: Dr. Daniel Crowther and Dr. Keith Vass

2. Outline Introduction Aims and Objectives Review of Literature Methods Results Conclusions Summary

3. Introduction Colorectal Cancer is the second leading cause of cancer death Gene expression profiling using microarrays is one of the effective methods to infer molecular mechanisms The main purpose: To identify gene expression modules that conclusively exist in colorectal cancer The outcome: Emphatically identified functional gene expression modules and mapped with other studies

4. Aims and Objectives To identify gene expression modules that are common to many colorectal cancer microarray datasets using three different methods To find biological relevance of the modules using BiNGO, NetAffix and KEGG pathway online tools Mapping with other studies: Predict probability of tumour relapse in colorectal cancer patients Predict probability of survival in breast cancer patients

5. Review of Literature Ruan et al. 1 successfully identified co-expression modules enriched with cancer related Gene Ontology categories from colon cancer microarray data Staub et al. 2 successfully classified various cancer patient populations using the genes present in WIPF1 co-expression module

6. Methods and Materials Data Collection from public databases Analytical methods Normalisation Multi-cluster comparison using Eigen Decomposition Correlation Similarity Pooled Co-expression Network – Clustering by Eigen Decomposition Correlation Similarity Pooled Co-expression Network – Clustering by Cytoscape – MCODE plug-in Analysis of the modules using BiNGO ,NetAffix and KEGG Pathway online tools Kaplan-Meier Survival Analysis

7. Results and Discussion The nodes (probe sets) are shown as red circles and the edges are shown as blue lines. The modules are named after their MCODE cluster ranks The number of nodes present in the modules 1 to 4 is 40, 32, 24 and 17 respectively Visualisation of gene expression modules in Cytoscape

8. Results and Discussion Gene Ontology categories overrepresented in the modules are shown in darker shades This module is highly enriched for genes present in cell cycle pathway BiNGO Gene Ontology enrichment results visualised in Cytoscape

9. Results and Discussion Kaplan-Meier Survival Analysis Plot showing probability of tumour relapse over time: predictability using MCODE cluster No.8 (RAS signalling pathway) as classifier on Wang et al. 3 Colorectal Cancer dataset No of samples = 74 Class Samples Relapse Cluster 1 42 15 Cluster 2 32 16 p-value = 0.3817 Red curve = Cluster 1 (under expressed) Blue curve = Cluster 2 (over expressed)

10. Result and Discussion No of samples = 235 Class Samples Death Class=0 47 9 Class=1 94 29 Class=2 94 17 p-value = 0.0168 Blue curve = Class 1 (over expressed) Black curve = Class 2 (under expressed) Kaplan-Meier Survival Analysis Plot showing probability of survival over time: predictability using MCODE cluster No.1 (Focal adhesion pathway) as classifier on GSE 3494 Breast Cancer dataset

11. Conclusions Emphatically identified gene expression modules that regularly occur in colorectal cancer Conclusively found functional significance of the modules Mapped with other studies to show their biological relevance Suggestions for future work

12. Summary The main purpose Three different methods Results Conclusions

13. References 1. Ruan, X. G., Wang, J. L. and Li, J. G. (2006), "A network partition algorithm for mining gene functional modules of colon cancer from DNA microarray data", Genomics, proteomics & bioinformatics / Beijing Genomics Institute, vol. 4, no. 4, pp. 245-252 Staub, E., Groene, J., Heinze, M., Mennerich, D., Roepcke, S., Klaman, I., Hinzmann, B., Castanos-Velez, E., Pilarsky, C., Mann, B., Brummendorf, T., Weber, B., Buhr, H. J. and Rosenthal, A. (2009), "An expression module of WIPF1-coexpressed genes identifies patients with favourable prognosis in three tumour types", Journal of Molecular Medicine (Berlin, Germany), vol. 87, no. 6, pp. 633-644. Wang, Y., Jatkoe, T., Zhang, Y., Mutch, M. G., Talantov, D., Jiang, J., McLeod, H. L. and Atkins, D. (2004), "Gene expression profiles and molecular markers to predict recurrence of Dukes' B colon cancer", Journal of clinical oncology : official journal of the American Society of Clinical Oncology, vol. 22, no. 9, pp. 1564-1571.

14. Acknowledgements Thanks to Translational Medicine Research Collaboration (TMRC), Dundee Thanks to Scottish Bioinformatics Forum Thank you