Which htseq-count mode to use for handling overlaps during RNA-Seq quantitation

Question

[skhan]

For counting gene expression levels (paired end stranded RNA-Seq data from mouse) using htseq-count, is there a universally accepted mode to use from the following list?

--mode union
--mode intersection-strict
--mode intersection-nonempty

To me, "union" seems like the appropriate choice but the wiki here suggests "intersection-strict" is best.

Answer 1

[lieven.sterck]

In most cases: yes, intersect-strict is the most appropriate one.

take a look here to see what the difference in behaviour of htseq is for the different modes. Depending on what you want to know/count you can switch to any of the other modes