Dear all,

I've started working with the fantastic csaw package.

I was searching for options or tweaks to take strand into account, since we have some strand-specific IP library preps performed.

Is there a way to incorporate this into the analysis?

Thank you in advance,

Ioannis

***Edit

We are analyzing stranded library preps such as: MeDIP-Seq, various RBP-IP protocols such as CLIP-Seq (some are stranded), etc.

***/Edit

It depends on what you mean by "strand-aware". Currently, csaw is slightly strand-aware in that directional read extension behaves differently for reads mapped to different strands. Other than that, though, there's not much distinction made between strands.

I've have to know more about the nature of the strand specificity in your experiment, in order to give some better advice. I don't think I've ever dealt with strand-specific IP's before, given that the pulled-down DNA is double-stranded to start with (unless you've got stuff like ChIP-Exo data).

---

Edit: Given that strand-specific counting seems to be pretty useful, I've added a forward option to the readParam object. This'll tell windowCounts (or regionCounts) whether or not to extract and count reads on the forward, reverse or both strands.

I've also added a strandedCounts wrapper function to easily get both forward- and reverse-strand counts for genomic bins or regions. Word of warning, though; the downstream functions in csaw that work with GRanges are not designed to be strand-aware, so some extra work will be necessary if you want, e.g., sensible strand-specific clustering. Check out ?strandedCounts for examples.

These updates should be active in the 1.1.19 build on BioC-devel; give it a day or so to go through.