VBZ fast5 results in failure to read fast5

[mattdmem]

Newer versions of MinKNOW now compress fast5 files with VBZ compression which causes strique.py count to fail.

Nanopore have provided an updated hd5 plugin to handle VBZ, however I'm struggling to get this to work with STRique

https://github.com/nanoporetech/vbz_compression

I've successfully used the tool they provide in the fast5 api (compress_fast5) to convert back to gzip which fixes the issue so it's definitely the switch to VBZ compression.

Any ideas (apart from changing all fast5 to gzip)?

[giesselmann]

could you try setting the following environment variable to point hdf5 to the compression (libvbz_hdf_plugin.so.1.x.y):
export HDF5_PLUGIN_PATH=/somepath/lib

[sxz-ivan]

I had the same problem (required filter 'vbz' is not registered), and solved it by installing the following after referring to jts/nanopolish#945 (comment):
pip3 install pycparser-2.21-py2.py3-none-any.whl
pip3 install cffi-1.15.0-cp36-cp36m-manylinux_2_5_x86_64.manylinux1_x86_64.whl
pip3 install pyvbz-1.0.1-cp36-cp36m-linux_x86_64.whl
dpkg -i ont-vbz-hdf-plugin_1.0.2-1.bionic_amd64.deb