Hi Dale,

Can I ask what platform/operating system you are running on?

We are aware of some issues on ubnutu bionic where HDF doesnt load the plugin automatically.

Can you also try adding:
export HDF5_PLUGIN_PATH=/usr/local/hdf5/lib/plugin to your shell before running the basecaller (in the same shell)

Thanks for your reply.
I try to add new HDF5_PLUGIN_PATH and it works without error information.

My platform is Linux version 3.10.0-693.11.1.el7.x86_64.

My another quesition is who can I decompress the vbz-compressed fast5？ Can I get the same raw fast5 files after decompression？

Best~

You can - we are working on some tools in our fast5 api to do this for you, however in the mean time you could use h5repack to remove the vbz compression, something like:

>  h5repack -f GZIP=1 input.fast5 output.fast5

should repack using gzip not vbz.

https://github.com/nanoporetech/ont_fast5_api has tools to repack fast5 files with or without vbz

Hi there-
I've been running into this issue as well using nanopolish polya on an M1 macbook. I ran
export HDF5_PLUGIN_PATH=/usr/local/hdf5/lib/plugin:$HDF_PLUGIN_PATH
in a terminal window followed by
nanopolish polya --threads=8 --reads=/path/to/joined.fastq --bam=/path/to/joined.sorted.bam --genome=/path/to/reference.fa > /path/to/polya_results.tsv
in the same window, which resulted in the following output:

The fast5 file is compressed with VBZ but the required plugin is not loaded. Please read the instructions here: #5

Sorry if I'm overlooking something obvious here, but I'd appreciate any input you might have on this issue. Thanks!

It sounds like you may have hit an issue with the vbz plugin on M1 platforms

Are you are running python in a rosetta environment?

A short term solution would be to build the VBZ plugin yourself on M1 - I will look to get M1 support for vbz rolled out in a future release.

@mmullistb can you try with the M1 Mac build ont-vbz-hdf-plugin-1.0.1-Darwin-arm64.tar.gz from the release page here.

thanks @jorj1988 and @iiSeymour - I downloaded the M1 mac build and added the directory containing the binary to my path:
export PATH=/path/to/ont-vbz-hdf-plugin-1.0.1-Darwin/bin:$PATH and confirmed with echo $PATH that it had been added.

I repeated the nanopolish polya command:
nanopolish polya --threads=8 --reads=/path/to/joined.fastq --bam=/path/to/joined.sorted.bam --genome=/path/to/reference.fa > /path/to/polya_results.tsv

The output was the following:

The line "The fast5 file is compressed with VBZ but the required plugin is not loaded. Please read the instructions here: #5" suggests that nanopolish is still not able to access the VBZ plugin, right?

@mmullistb did you use PATH not HDF5_PLUGIN_PATH?

@iiSeymour haha yes, my mistake. It's working now when using HDF5_PLUGIN_PATH. Thanks!

Hi there,

I'm running nanopolish eventalign on some of my data and the error code I'm getting is directing me to this thread.

The fast5 file is compressed with VBZ but the required plugin is not loaded. Please read the instructions here: https://github.com/nanoporetech/vbz_compression/issues/5

I am suspicious of this error as eventalign runs smoothly when given the example data found in the guide

Fixes attempted

• I attempted the fix described above however my installation of hdf5 doesn't have a file or folder named plugin in the /usr/local/hdf5/lib/ directory, and thus there was no change in output.
• I attempted another fix described above to a similar lack of effect
• I repacked the fast5 files as per above, using /home/my_username/Apps/nanopolish/hdf5-1.8.14/tools/h5repack/h5repack -f GZIP=1 $i ../fast5rp/$i there were many warning messages stating that 'warning: dataset </read_ff33cbfc-f8d1-4139-bb06-6c05854ae7bf/Raw/Signal> cannot be read, user defined filter is not available' throughout this process. After reindexing using these repacked fast5 files the code now completes but the post run summary claims that all 9011 reads are 'bad fast5'. Which is interesting as I know that this file contains around 16000 reads and not 901.
  [post-run summary] total reads: 9011, unparseable: 0, qc fail: 0, could not calibrate: 0, no alignment: 0, bad fast5: 9011
• I ran this code on a different system with the exact same error codes
• I tried using the distribution of nanopolish available through apt and found a different set of errors that I investigated using gdb to find a 'no such file or directory' error which I can give more details on upon request.

I find it difficult to believe that my fast5 files are corrupted as I am able to complete basecalling and mapping without obvious errors seen when running fastQC. I think the program is right with its first error message and that it is decompressing the files incorrectly, hopefully you can help me in configuring my installation to appropriately process these files.

Thanks,

Finnlay Lambert

Hi @finn-rpl ,

what version of the vbz plugin do you have installed? What OS (+ version) are you running?

Thanks,

• George

Hi George,

I'm running Ubuntu 20.04, but the issue is solved and the error was mine.

When I tried the fix detailed here, the link downloaded the mac version of the plugin rather than directing me to the releases page.

After installing the correct version the export HDF5_PLUGIN_PATH=/usr/local/hdf5/lib/plugin fix worked for me first try.

Thanks for your assistance, I only realised there were multiple distributions after you asked me which was the appropriate version.

Finnlay Lambert

Perfect - Thanks for updating me!

Hello I am using nanoseq pipeline to do M6A analysis. I am using HPC cluster for analysis. I installed https://github.com/nanoporetech/vbz_compression/releases/download/v1.0.1/ont-vbz-hdf-plugin-1.0.1-Linux-x86_64.tar.gz in my conda virtual env. I used following command to run the pipeline nextflow run

nf-core/nanoseq --input samplesheet.csv --protocol directRNA --skip_demultiplexing -profile singularity -c vbz.config -r 3.1.0 --skip_fusion_analysis --skip_differential_analysis..

My custom config file has

process {
    withName: NANOPOLISH_INDEX_EVENTALIGN {
        container = 'https://depot.galaxyproject.org/singularity/nanopolish:0.14.0--h773013f_3'
    }
}

env {
    HDF5_PLUGIN_PATH = '/data/user/home/mbansal/.conda/envs/ont-vbz/hdf5/lib/plugin'
}

but I am keep getting error

[readdb] indexing fast5
[readdb] num reads: 427756, num reads with path to fast5: 427756
The fast5 file is compressed with VBZ but the required plugin is not loaded. Please read the instructions here: https://github.com/nanoporetech/vbz_compression/issues/5
HDF5-DIAG: Error detected in HDF5 (1.12.2) thread 1:
  #000: H5D.c line 276 in H5Dopen2(): invalid location identifier
    major: Invalid arguments to routine
    minor: Inappropriate type
HDF5-DIAG: Error detected in HDF5 (1.12.2) thread 1:
  #000: H5G.c line 502 in H5Gopen2(): invalid location identifier
    major: Invalid arguments to routine
    minor: Inappropriate type

Hi @vetmohit89 ,

Can you confirm the architecture of the system you are running on? What is the content of the directory /data/user/home/mbansal/.conda/envs/ont-vbz/hdf5/lib/plugin ?

Have you tried running the same experiment outside of your container?

• George