while running ONT(nanopore) bioinforamtics pipeline, fasta files are not generating.

Question

[priyankamehta.1811]

Hello, I am trying to run the ONT pipeline for generating fasta for covid and working on the HPC server. while running a test set on my system, I was able to generate fasta files along with vcf and bam files using the minion protocol. however, while running a larger set on HPC, I am only getting the vcf and bam files but no fasta files are generating. I have tried to re-install the artic-ncov2019 environment in Anaconda. When my base-called files are run on a DGX server, fasta files and all other files are formed.

However, in the HPC server, we have a GPU node but running minion on that doesn't generate fasta.

I am utterly confused as it's my first time working on the server. But so far as I know the type of server doesn't matter for generating fasta.

Can anyone suggest what I can do?

Thanks

[GenoMax]

Sounds like last part of the pipeline may not be running properly. It should be easy to change the fastq files (if you see them) into fasta by using reformat.sh from BBMap suite.

reformat.sh in=your.fastq out=out.fa

[chrys]

This might be a very obvious question but did you install the GPU version of Guppy on the GPU Server?

Answer 1

[colindaven]

I think you need to clear up the terminology you're using a bit

• MinKNOW is the control software for Minion. I think this is what you mean when you say Minion. It can now be used for alignment, which I think is what you've been doing to generate BAM and VCF. As a side note, you can generate FASTQ from BAM using many different toolkits, including recent samtools versions.
• Guppy is the basecaller for FAST5 files/signals/reads. You can use this, either as a CPU or GPU version, to basecall your FAST5 files/ directories and create FASTQ files. You probably also have barcodes so will need to look at this. Another tool for assessing barcodes is https://github.com/rrwick/Porechop

Specifically to your question - no, the type of server does not matter. You might be better off using Miniconda rather than the whole anaconda. Your sysadmins might be able to help installing system packages of Guppy which might be needed to properly pick up the GPU on the GPU server. We have to use Singularity to get Guppy working on Ubuntu 20.04 for example, since packages only exist for 16.04 and 18.04 AFAIK.

I hope that helps.