search

HKU-BAL / Clair3

Clair3 - Symphonizing pileup and full-alignment for high-performance long-read variant calling
244 stars 26 forks source link

SNP calling from bisulfite converted ONT #256

Closed weishwu closed 10 months ago

weishwu commented 10 months ago

We are planning to use the Twist protocol to do targeted ONT sequencing. The protocol includes an enzymatic conversion step to convert all unmethylated Cs to Ts, so the analysis will be similar to Illumina BS-Seq. Our purpose is to do allele-specific methylation so we have to first call SNPs and then use them to phase methylation to the two alleles. Is there a way to use Clair3 for this type of SNP calling? If not, does anyone know of any other tool that I can try? I know there are multiple BS SNP callers for Illumina, but I'm not sure if they are applicable for ONT. Besides, my experience is that the accuracy of BS SNP calling is not that good.

Ideally, we should use adaptive sampling to do the targeted sequencing so that we can keep the DNA native. However, AS right now does not provide high enough fold enrichment and we are restricted by funding.

aquaskyline commented 10 months ago

I think it's doable, but instead of using all SNPs, please maybe choose only the top n% SNPs for your phasing purpose. I believe Clair3 can still identify most of the C->T, but the model would still consider the excessive number of C->T abnormal and thus lowering their scores.

weishwu commented 10 months ago

@aquaskyline Thanks!