Arthur Rand

@faulk-lab, Ok. I pushed a change that will log when this happens instead of firing an assert. I'm going to try and find an example that exposes the problem on...

Great. I'm doing some sanity-check experiments as well. 4 CpGs seems to be the "standard" for BS-Seq tools, but from what I've seen (and I might be wrong) there is...

I think you want an option to drop reads with >X% filtered calls in the window.

Here's a build with `--drop-zeros` and `--max-filtered-positions`. I found that requiring that a read has at least 3 passing calls (`--max-filtered-positions 1`, `--num-positions 4`) improves the resolution of the output....

Nope, that's a bug 🐛 . I'll fix it.

Should be fixed on the branch. Here's a build. [modkit_dev7acefc4_centos7_x86_64.tar.gz](https://github.com/nanoporetech/modkit/files/14621634/modkit_dev7acefc4_centos7_x86_64.tar.gz)

Hello @faulk-lab, find anything interesting? I'm prioritizing which features to clean up and get into the next release. I'll also settle for "the entropy calculations make sense, but I haven't...

No problem at all. I like the idea of increasing the `--num-positions` parameter, using 4 feels like a "short-read" legacy artifact. Glad it's working, I'll push to get it into...

I can add the calculation over input regions, it's easy to implement. However, one wrinkle is what do you do if none of your reads span the entire region? Do...

Interesting, for the windowing I took a queue from [dnmtools](https://dnmtools.readthedocs.io/en/latest/entropy/) actually. I think regions makes sense, of course as you mention, you'll have to keep the length of the region...