Aims

2. The Pan-Cancer initiative provides a model for large-scale collaborative analysis as well as data sharing, bringing together over 250 collaborators from ~30 institutions working together on over 60 projects analyzing the same dataset.

3. As of Jul. 24th 2013, TCGA had mapped molecular patterns across 7,992 total cases representing 27 tumor types.

4. Aims • To generate, quality control, merge, analyze and interpret molecular profiles at the DNA, RNA, protein and epigenetic levels for hundreds of clinical tumors representing various tumor types • To increase the statistical power to detect functional genomic determinants of disease • To identify commonalities and differences across tumor types.

5. Cancers of different organs have many shared features, while cancers from the same organ or tissue are often quite distinct. • Shared molecular patterns will enable etiologic and therapeutic discoveries in one disease that can be applied to another. • integrative interpretation of the data will help identify how the consequences of mutations vary across tissues, with important therapeutic implications.

6. Use of the data enables the identification of general trends, including common pathways (bottom left), revealing master regulatory hubs activated (red) or deactivated (blue) across different tissue types.

7. The Synapse system created by Sage Bionetworks. • the use of the Synapse software platform to share and evolve data, analysis and results among the Pan-Cancer Working Group. • "This beautifully organized data repository is now available for scientists around the world to use to go beyond these initial analyses and discover even more about cancer“

8. Challenges • Integrate data generated on different platforms or updates of the same platform • Batch effects • The nature and quality of available clinical data vary widely by cancer type • Each tumor type has its own system. Not easily comparable across different tumor types. • Statistically speaking, cross-cancer comparison does not lead to increased false-negative rates for discovery or false-positive rates

10. miRNA, distribution… • Clinical tumor analysis • Subtype/Subclass… • Big data integration…

11. A hierarchical classification of 3,299 TCGA tumors from 12 cancer types • Oncogenic signature classes • Targetable functional events

12. Bipartite network modularity for recurrent genomic alterations

13. Discussion • provides a systematic approach for integrating large amounts of molecular data in a way that reduces its complexity (noise) and increases its biological and clinical interpretability (signal). • systematically derive signatures of functional alterations • class-specific/personalized combination therapy

14. Douglas Hanahan • Professor, Biochemistry • Swiss Institute for Experimental Cancer Research (Director) • UCSF Comprehensive Cancer Center (Program leader, Mouse Models of Cancer Program), UCSF Diabetes Center (Member) • Pancreatic cancer

15. Summary • Analyze gene expression profiles from 1,290 CRC tumors using consensus-based unsupervised clustering. • Six clinically relevant CRC subtypes • shows differing degrees of ‘stemness’ and Wntsignaling • Disease-free survival (DFS) • Drug response • Subtype-specific gene signatures

19. Methods • NMF consensus clustering

20. The end • Thanks!