Variant is not detected in amplicon sequencing data

[keremozdel]

Hello,

I am using Clair3 for variant calling on amplicon sequencing data. These reads were basecalled using Dorado (dna_r10.4.1_e8.2_400bps_sup@v4.3.0 model). For Clair3, I am using the r1041_e82_400bps_hac_v420 model (from Rerio). I have set the parameters to --var_pct_full=1 --ref_pct_full=1 --var_pct_phasing=1. However, Clair3 does not detect a variant at a specific position, which is visible in IGV. I also tried running Clair3 with r1041_e82_400bps_hac_v430 and r1041_e82_400bps_sup_v430 models, but the result was the same. Furthermore, I observed a decrease in recall when I used v430 models. I attached the IGV image of the variant in question:

Do you have any idea what might be the reason?

Thank you for your assistance.

[aquaskyline]

Positive strand reads are two times more than those of negative strands. Is it expected?

[keremozdel]

It is not expected. I suspect the primer efficiency varies between the primers, and I am wondering whether this affects the accuracy of variant calling, even though there are sufficient reads for both strands?

[aquaskyline]

In all available Clair3 models to date, a balance between two strands is required. So my guess is that Clair3 rejected your case primarily because of strand bias.

[keremozdel]

This makes sense. I will investigate the reasons behind that strand bias. Thank you very much for your help.