OncoLand Case Study - Differential Expression within and between Tumor Types
From Array Suite Wiki
Esophageal tumor subtypes closely resemble head/neck and stomach cancers
A recent paper used TCGA data to examine molecular subtypes of esophageal cancer and compared them with other tumor types.
We can use TCGALand to verify and explore these results, using SampleSets and Integration Analyses.
Study Background - Esophageal Cancer
Identify and Group Samples within Land
The default view within Lands is the sample distribution view. This view will display all samples within the land and organize the graph based on metadata. Land metadata can be used to filter samples of interest and group samples within the view.
Create a sample set using defined filters and grouping
Once samples of interest are found, they can be used to generate sample sets. These sample sets provide an easy way to quickly filter back to these samples in the future.
Clinical Association and Survival
Once samples of interest are identified, users can quickly scan clinical and numeric variables that are associated with the grouping as well as examine differences in survival between groupings.
Performing Integration Analysis
Sample Groupings can be queried using the numerous Land Data types, such as Expression (microarray, RNA-seq, proteomics, miRNA). For expression analysis, this takes user-defined groupings and performs Kruskal-Wallis tests to identify differential expression.
Generating HeatMap Views of Differential Expression
Once integration analysis is complete, users can sort through differential expression and visualize expression of genes dynamically within Land.
Comparing gene expression with other tumor types
Subtypes of esophageal cancer are derived from different regions of the esophagus. To see how these subtypes compare to tumors derived from nearby locations, namely head/neck and stomach, these tumor types are simply added to our views and updated.