dorado v0.9.0 does not remove adapter sequences from Ultra-long data

[is01]

Hello,
I basecalled ultra-long data with dorado v0.9.0, it seems that ~90 bases of adapter are left.
The adapter trimming worked in v0.8.2.
Is there an additional option required to trim the adapter sequence of the ultra long kit?

Run environment:

• Dorado version: v0.9.0
• Model: [email protected]
• Dorado command: dorado basecaller -r --reference $fasta --modified-bases 5mC_5hmC $model $pod5 > out.bam

[malton-ont]

Hi @is01,

Thanks for reporting this. dorado 0.9.0 is supposed to select the appropriate adapters and primers to trim based on the sequencing kit detected from the pod5 file. It looks like the ULK kits were missed in the look-up - we'll get a fix out for this asap.

[malton-ont]

Hi @is01,

This issue should be resolved in dorado 0.9.1. If it persists, please re-open this ticket.

[is01]

@malton-ont Thank you for your quick response. I will try new version.
thanks.

[is01]

I basecalled the ultra-long data with dorado 0.9.1, but the reads still have adapters/primers.
It seems that the RAD primers (transposase sequence) were not removed.
Thanks for help.

[malton-ont]

@is01,

Are you able to share a pod5 file containing a few example reads?

[is01]

@malton-ont,
I'm sorry I can't share the pod5 file, but I found that adapter trimming works fine by specifying a custom primer file.
I ran the following command (dorado 0.9.1):
dorado basecaller -r --reference $fasta --modified-bases 5mC_5hmC --primer-sequences primer.fa $model $pod5 > out.bam

Custom primer file (primer.fa):

>RAD_front type=primer kits=SQK-ULK114
GCTTGGGTGTTTAACCGTTTTCGCATTTATCGTGAAACGCTTTCGCGTTTTTCGTGCGCCGCTTCA

Is it possible that the above sequence is not searched for in the ULK kit?

[malton-ont]

@is01,

Apologies! dorado 0.9.1 added this kit to the adapter search, but is still missing it for the primers. We'll address that in the next release. In the mean time, using the custom primer trimming as you have is the appropriate solution. The other option is to add a dorado trim step after the basecalling and tell it to use the RAD114 kit (which has the same primer)

dorado trim --sequencing-kit SQK-RAD114 out.bam > trimmed.bam

[is01]

@malton-ont,

Thanks for providing the solution. I look forward to the next release.