These are chat archives for dereneaton/ipyrad

10th
Jan 2016
Deren Eaton
@dereneaton
Jan 10 2016 19:05
I suppose the filter() function could shunt the filtered loci to a file. It's useful for debugging to make sure the filter is working how we want it. But otherwise no I don't think user's typically use the excluded loci for anything.
Deren Eaton
@dereneaton
Jan 10 2016 21:01
I've changed my mind about "population" being an attribute of Samples and a @property of Assembly. It makes more sense that a Sample is its own independent thing and that different Assemblies will group Samples together differently. So a Sample passed between Assembly objects should not be expected to retain its population information.
Isaac Overcast
@isaacovercast
Jan 10 2016 21:22
agreed. luckily that's how i coded it B-)
I'm in sequence filtering hell. All filters are almost done, just have to make sure they all agree on what they're read/writing.
Deren Eaton
@dereneaton
Jan 10 2016 21:41
Ok, I got caught up cleaning up some stuff in demultiplexing that is bothering me.
Isaac Overcast
@isaacovercast
Jan 10 2016 21:58
You ever do that thing where your like "yeah, i'm gonna do this really smart", but then all that happens is you do it really complicated and it breaks alot? My first pass at filtering was that way, but now i figured out the simple easy way.
Deren Eaton
@dereneaton
Jan 10 2016 22:39
ha, yeah. hopefully we're not going too far down the complicated path.
Deren Eaton
@dereneaton
Jan 10 2016 23:06
The most recent setup.py asked me for my username and password when querying github for the tag... That won't happen to other people running pip install will it?
Isaac Overcast
@isaacovercast
Jan 10 2016 23:13
No because the way it is now is hackish and i want to fix it, i'm thinking about pulling the 'magic' code out of setup.py and just having a devs only util, just a little py script so if we want to update the version we can say "./ipyradupdateversion 0.2.0" and it'll do all the magic of updating init, adding a git tag, pushing the tag, committing init and pushing. It's still on my list, but i'm prioritizing step7 right now.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:23
Query: Do you think folks want the full vcf with genotype likelihoods or just the snps
Deren Eaton
@dereneaton
Jan 10 2016 23:32
oh, sounds cool.
I think 95% of users just want the SNPs. As it is now any genotype likelihood <95% is called N.
So we know all called bases are above 95
we throw away that info at step5, but like I said, we could keep it. But I wouldn't prioritize it.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:42
CHeck.. For the purposes of the .loci snpsites line would you count 2 different singletons at a base as an autapomorphy?
Deren Eaton
@dereneaton
Jan 10 2016 23:42
yeah
man, I can't find any of the old pairddrad data sets I used to test on
I must have deleted them to make space. I'll find one online.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:43
Ok good. I'm protecting against the case where there are up to 3 singletons and still calling it parsimony uninformative...
Deren Eaton
@dereneaton
Jan 10 2016 23:44
The pairgbs data set I'm testing on right now is such garbage that all reads are either discarded or merged.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:44
D::::::
Deren Eaton
@dereneaton
Jan 10 2016 23:44
I'll send it to you soon to play with, tho. I'm making a notebook with some tips on how to filter it.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:45
That'd be super useful, messy data is best to test with. The sim data is so clean I'm having a hard time telling if the filters work.
Isaac Overcast
@isaacovercast
Jan 10 2016 23:52
Even after all this time i am amazed at the kinds of stupid mistakes I can make. The amount of time I've spent spent searching for stupid indexing errors, type-o's, etc, etc is fucking measurable... Just found another good stupid one lol.
Deren Eaton
@dereneaton
Jan 10 2016 23:54
lol, I've had many days spent similarly.