Closed lfearnley closed 2 years ago
Sorry for the slow response. We don't have a hard requirement for numpy v1.22.2 on the dev branch. pip
should, in this scenario, downgrade numpy to the pinned version used by the read_until_api
but (as you mention) pandas is preventing this.
I think the solution here is ensuring that pandas is set to ==1.3.?
. I have tried 1.3.1
and 1.3.3
which both install successfully. Please let me know how you get along with this
Hi @alexomics! Sorry about the delay - I've had a bit of a gap where I didn't want to touch the readfish install.
Setting pandas to 1.3 seems to be working ok in simulation, though I'm noticing a bit of a slowdown using guppy 6.0.6 relative to older versions of readfish - I think this might be a guppy issue though, as it's occuring on both a 2080Ti and an RTX A5000.
Hi all!
Hopefully an easy one - I'm following the readfish dev branch install instructions in a conda environment, and I'm seeing a dependency issue when running the commands as specified:
pip install git+https://github.com/nanoporetech/read_until_api@v3.0.0
installs and requiresnumpy==1.17.4
.pip install git+https://github.com/LooseLab/readfish@dev
installsnumpy==1.22.2
and generates an incompatibility warning.What's the cleanest way to resolve this? Will readfish run without issues on numpy 1.17.4/pandas 1.3.x?