venn

[Instructions]

venn
Create venn diagrams based on molecular features. Also generate plots and biomarker frequency reports.

The venn function allows for a large degree of customization of the thresholds used to determine biomarker status. However, the cost is that thresholds for each gene will have to be evaluated individually before generating the final output. The report can be looped over all the TCGA datasets and summarized with the caveat that thresholds may differ between datasets.

While the venn diagram can handle up to four genes, the report can utilize up to 9 genes. The order of the genes in the HUGO box corresponds to the gene number for the threshold cutoffs.

Thresholds for gene expression and CNV are expressed as > N or < N. eg < 10. If you do not want to utilize a given threshold, set it to an unrealistic number. eg. gene expression > 20. ( range is typically 0-14 )

plot ALL TCGA datasets : when checked, loop the analysis over each TCGA dataset and create a summary table. (default unchecked)
Summary table only segment summary table by : when checked, further segment the report by the drop down selection. (default unchecked)
choices:
first gene : segment the report by the values of gene1
first and second gene : divide the groups by the values for gene1 AND gene2
metadata : when METADATA selected, use the assigned metavalue to define the groups.

example:
put esr1 erbb2 rb1 into the HUGO box
set datasources for expression, cnv, mutation to TCGA-BRCA (RSEM for expression)
press submit

check the cutoff values from the scatter plots and histograms produced. The red or blue color shows which samples will be included in the cutoffs. For gene expression, ESR1 looks bimodal with a cutoff ~10. ERBB2 is also bimodal but the cutoff is ~15. For RB1 we do not want to use expression to define status, only mutation and CNV status, so we can set the cutoff to >20.
Modify the gene expression cutoffs to >10 for gene1, >15 for gene2, and >20 for gene3.

For CNV, we are not going to use ESR1 so set gene1 > 3. For ERBB2, set gene2 > 0.75 as the exp vs. cnv plot shows most amps >0.75 have high expression. For RB1 we are interested in deletions so set gene3 <-0.75

resubmit

Development:

venn summary table. When segmenting by metadata, can further sub-segment by adding additional meta values to the HUGO box using the META-NNN value. (eg. subsegment by both race and subtype)