These are chat archives for dereneaton/ipyrad

24th
Feb 2018
toczydlowski
@toczydlowski
Feb 24 2018 16:54
@isaacovercast Thanks Isaac. I don't see a logical explanation (yet) for how oversplitting would yield this specific pattern but I'll keep that in mind. I am wondering if it has to do with Stacks not reverse complimenting and/or allowing indels. Is there a way to turn reverse complimenting off in ipyrad to try to test part of this? ipyrad won't assemble loci without indels correct? It just throws loci that have indels in them out at the end in step7 if you ask it to (maxindels=0).
Isaac Overcast
@isaacovercast
Feb 24 2018 18:04
@toczydlowski revcomp and indels could both be contributing, true. You can effectively disable revcomp in ipyrad by running gbs as 'rad' data. Also, yes if you set maxindels to 0 it'll throw out reads at step 7, but this won't be comparable to the stacks results because ipyrad will still identify clusters with indels, whereas stacks will split true loci that contain indels into 2 artificial loci.
toczydlowski
@toczydlowski
Feb 24 2018 20:18
@isaacovercast Ahh, interesting re: how to disable revcomp. That makes sense, and I assume the data type is the only param I would need to change? I might try that and see how results compare. And remind me - is there a stat somewhere that will tell me how many reads were reversed by revcomp? Thanks for confirmation on indel behavior.
Isaac Overcast
@isaacovercast
Feb 24 2018 23:07
@toczydlowski Mm, we don't keep track of how many reads clustered revcomp for gbs.