Dorado duplex reads' mean quality are lower than simplex reads

[ErminZ]

Possible reasons for duplex reads' quality are low

Hello, thank you for increasing the read quality! We always observe higher read mean quality on duplex than simplex reads, especially simplex paired reads. However, recently one sample has the opposite shown in the picture. Could you explain why duplex reads' quality is lower than simplex reads?

The mean quality of duplex reads is lower than simplex shown below:

Here are two duplex reads and their simplex paired reads examples:

Would you explain possible reasons why the duplex reads' quality is low? Please let me know if you have any questions. Thank you!

Sample information:

1. PCR-targeted sequencing: two amplicons (500bp wild type and 2.2kb insertion into wild type);
2. Hybridization to enrich the reads only contains the 1.7kb insertion;
3. PCR again amplifies the enriched 2.2kb reads with the targeted insertion.
4. The flow cell is overloaded due to the inaccurate read length evaluation before sequencing.

Run environment:

• Dorado version: V0.7.1 docker genomicpariscentre/dorado:0.7.1

• Dorado command: dorado duplex sup ${pod5_directory} > ${sample_id}.bam --min-qscore 10

  dorado summary ${sample_id}.bam > ${sample_id}_dorado_summary.tsv

• Operating system: Linux
• Hardware (CPUs, Memory, GPUs): AWS EC2 p3 8 GPUs instance
• Source data type (e.g., pod5 or fast5 - please note we always recommend converting to pod5 for optimal basecalling performance): pod5
• Source data location (on device or networked drive - NFS, etc.):
• Details about data (flow cell, kit, read lengths, number of reads, total dataset size in MB/GB/TB): flow cell: kit: SQK-LKS114 Read length: total dataset:

  
  Simplex reads basecalled: 1,396,6178
  [info] > Simplex reads filtered: 690,388

[info] > Duplex reads basecalled: 4,882,572 [info] > Duplex reads filtered: 510,211 Duplex rate: 49.04696% Basecalled @ Bases/s: 3.975039e+05

- Dataset to reproduce, if applicable (small subset of data to share as a pod5 to reproduce the issue):

## Logs

* Please provide output trace of dorado (run dorado with -v, or -vv on a small subset)

[vellamike]

Hello @ErminZ

I’m quite puzzled by the very low accuracy you’re seeing—it’s unexpected and strange, especially given that duplex reads typically outperform simplex. It’s unusual to see the opposite in this case. Are these Q scores predicted or aligned?

That being said, with your approach of sequencing duplex with amplicons, there’s no guarantee that the following strand will be the complement of the first one. Given that duplex relies on correctly pairing template and complement strands, this could lead to the issues you’re observing.

It probably makes more sense to use simplex for your application.

[ErminZ]

Thank you for the reply! It is very helpful.

Maybe only simplex reads are the best choice for PCR-targeted samples with >80% of reads being duplicates? Would you tell more about whether there are factors other than sequencing location, start/end time, and read sequences and length that influence the duplex paring? Thank you!

The Q scores are from the dorado summary function. I guess it is predicted?