Has anyone compared Dorado 0.6.0 to Guppy 6 on ONT R10.4.1 data for downstream variant calling? On an A100, Dorado increased mean read Q by about 2.3 and throughput by about 30% for me, but small indel performance in low-complexity regions is inconsistent between PEPPER-Margin-DeepVariant and Medaka — curious what patterns others are seeing.
I found that while Dorado increased my mean read Q, I’ve seen similar inconsistencies in small indel performance with low-complexity regions… It could be worth checking how your low-complexity settings are configured in PEPPER-Margin-DeepVariant. Have you tried adjusting those parameters?
And it’s interesting how Dorado boosts mean read Q but then leaves us feeling like we’re chasing a mirage with those indels! Have you tried adjusting any parameters in PEPPER-Margin-DeepVariant? I’m curious if that helps with the consistency.
I found that tweaking the parameters in PEPPER-Margin-DeepVariant did help to stabilize my indel calls a bit. @lisa_garcia87, have you experimented with different settings to see if that might enhance your results?