Currently recommended Oxford Nanopore data basecalling and preprocessing pipeline for de novo assembly applications

Question

[Denis]

Hi there,

We are just launching ONT data analysis and looking for an up-to-date tutorial containing pipeline for ONT data basecalling and preprocessing (e.g. adapter and quality trimming etc.) for de novo assembly applications. We are looking for the tools with command line interface (without GUI).

Many thanks, Denis

Answer 1

[lieven.sterck]

why not use the tool(s) provided by ONT itself, namely dorado (basecalling, demultiplxing, adapter, ..) ?

They provided decent info on how to use it.

for the quality filtering you can have a look at seqkit , and use nanoplot as a fastqc alternative (there are others available as well though)

[Denis]

Thank you! As i understand, the Dorado tool is applied for basecalling. But i'm also wondering which preprocessing steps and tools i should use before *de novo* assembly itself.

[lieven.sterck]

ah, not that much actually :-)

do a decent basecalling, perhaps some quality filtering, certainly adapter removal but that should be about it.

for assembly purposes , that adapter removal will be most critical, the quality filtering should more or less resolve itself during the assembly process, but it won't hurt to do it in advance (using mild settings not be to stringent here)

[Denis]

Thank you for nanoplot mentioning for data QC! Perhaps common Illumina short reads tools are suitable for quality trimming of ONT data. But it seems there should be some specific in adapter removal step intrinsic only for Nanopore technology. May it will be safer to use tools designed for ONT in adapter clipping step?

[Denis]

Besides, i'm wondering if it is proven now that Dorado the most accurate basecaller for ONT data. In particular in comparison with Guppy.

[GenoMax]

Yes dorado is the current basecaller for ONT data. It is actively supported and regularly updated by ONT.

certainly adapter removal but that should be about it.

dorado by default removes barcodes, adapters and primers. It will also demultiplex the data if you are using barcodes. It works efficiently on a GPU so you will preferably need access to one.

You can use fastplong (LINK) or PycoQC (LINK, you will need sequencing summary file from nanopore data folder) to look at QC (length, distribution of sizes etc). fastplong can also help with filtering/trimming etc.

[Asan Emirsale]

the dorado does remove adapters for only 10.4.1 etc data. If you have data from r9.4.1, you should activate ---trim flag.

I suggest also looking toward performing duplex basecalling. Given 'sup' models used in analysis you may obtain much better quality for de novo assemble.

Best regards, Asan

[Denis]

Thank you! A lots of important information in your comment!