I am wondering how the analysis is done for the “context” page, dependencies enriched in neuroblastoma (Neuroblastoma DepMap Context Summary). What does the t-value and p-value mean here? Does highly significance mean that there is a dependency for the gene/compound in most of the Neuroblastoma cell lines unrelated to other entities? Or does it mean that it is highly selective, so specific for neuroblastoma, and not a common essential gene? How can we extract all the “highly selective” genes in a certain disease?
Yes, you’re right. It appears we need to add some text describing the actual analysis that was run. In the mean time, I’ll describe it here:
In the portal when we refer to “enriched lineages” or “contexts”, what we’re saying is that mean of the gene effect for cell lines within that context appears to be different then the mean of the gene effect for cell lines outside of that context.
For each gene and context, we divide cell lines into an ‘in’ and an ‘out’ group based on the context and we compute a t-test to see if the resulting t-statistic is significant.
I should note that the p values are not corrected for multiple hypotheses. We report these same p-values on the gene page (where we show which contexts show enrichment for the given gene) and on the context page (where we show the inverse: which genes showed enrichment in this context). Based on your set of hypotheses, you may want to adjust your stringency on the p-value accordingly.
Hi, Phil. This was also what I was looking for. But then, how would you do the multiple testing correction? To my understanding, I do not know how many times a gene is tested between ‘in’ and ‘out’ group of cell lines and this impedes to set up the multiple testing correction.
