Hi all,
I am currently working with Nanostring data on different cancer types. From nSolver, I am able to get the raw count data as well as normalized data. However, since my project wants to focus on certain type of cancer (BR), I have filtered out DEG that are only detected in BR (list gene A). I am wondering if it is right to run GO enrichment only on list gene A instead?
I ran it with homo sapiens as the reference set, and only CC shown some results. But based on some reading, seems like by doing this, it will only produce biased result as these genes were already 'pre-enriched' from the genes panel used in Nanostring. I also ran the enrichment with genes' panel as the reference set, and only CC shown some results in which I guess we could say it is tally with the enrichment analysis above. However, I am also confused on how to proceed next with this little output I obtained.
I would appreciate if anyone can give some opinions on this. Thank you.
Thats right you should not filter out genes from DEG list. And it seems you have performed an over-representation analysis (ORA) on GO terms which is fine.Since you have DEG data, you could try gene set enrichment analysis (GSEA) which will provide you somehow unbiased results. If you are familiar with R, there are a buch of packages that one can use for enrichment analysis. Also GSEA got a version installable on most OS.
Hi, thank you so much for your reply.
I have seen GSEA being mentioned but as my analysis was obtained from Nanostring data (which is run based on specific set of panel genes), will it still considered as unbiased? What do you think?
Btw, luckily I am familiar with R so I will try it out first.