Representation of MNP

[lisosome]

I'm using Clair3 v1.2.0 to perform variant calling on ONT amplicon data. This is my Clair3 command:

/opt/bin/run_clair3.sh --threads=8 --platform="ont" --var_pct_full=1 --ref_pct_full=1 --var_pct_phasing=1 --min_mq=10 --enable_variant_calling_at_sequence_head_and_tail --haploid_sensitive --qual=15 --bam_fn=/orfeo/cephfs/scratch/burlo/nardone/minION_250331/reanalyses/STRC_analyses/25-703/2.ALIGNMENT/25-703.bam --ref_fn=/orfeo/cephfs/scratch/burlo/nardone/minION_250331/reanalyses/STRC_analyses/reference/STRC_ref.fasta --bed_fn=/orfeo/cephfs/scratch/burlo/nardone/minION_250331/reanalyses/STRC_analyses/25-703/3.VARIANT.CALLING/variant_calling_bed --model_path=/orfeo/LTS/burlo/LT_storage/shared/resources/clair3_model/r1041_e82_400bps_sup_v500 --output=${outdir}

In the resulting VCF I have two variants in two adjacent positions:

However, after sequencing the sample using WES, an indel is called at the same position, with the deletion of the reference bases (GT) and insertion of the alternative bases (AG). Inspecting the BAM from the ONT sequencing, this is the site:

What can be the reason why Clair3 is calling this variant as two distinct SNVs?

Best regards,
Giuseppe

[zhengzhenxian]

Hi @lisosome,

Based on the observation in IGV, I only see two SNPs. Therefore, the output from ClairS is consistent with the IGV. Two insertions followed by two deletions are typically represented as two SNPs in my understanding.

[lisosome]

Hi @zhengzhenxian ,

Thanks for the reply, and sorry for my late response.

The two variants are located at positions 43600609 (G>A) and 43600610 (T>G), respectively. The thing is, when looking at the annotation, the first is annotated as pathogenic, while the second as a synonymous variant.

That's not an issue per se. However, when looking up at the Clinvar record, the variant is registered as:

NC_000015.10:g.43600609_43600610delinsAG

So, wouldn't the variant be correctly represented as an MNP with the change GT>AG rather than two single variants?

I should add that the regions in which variants are called are haploid regions, due to a whole gene deletion occurring on the other allele.

Again, in the WES analysis, the variant is correctly called as GT>AG. So DRAGEN correctly classifies this variant as a MNP.

Best regards,
Giuseppe

[dariober]

I am seeing the same issue. Here's there is a MNP where the AGA in the reference is replaced by GAT in the sample (bottom line is the reference sequence):

However, clair3 gives me three SNPs:

PCHAS_01_v3	117614	.	A	G	31.62	PASS	P	GT:GQ:DP:AD:AF	1:31:334:0,334:1.0000
PCHAS_01_v3	117615	.	G	A	29.81	PASS	P	GT:GQ:DP:AD:AF	1:29:334:0,334:1.0000
PCHAS_01_v3	117616	.	A	T	29.15	PASS	P	GT:GQ:DP:AD:AF	1:29:334:0,334:1.0000

My organism is haploid and this is my command:

run_clair3.sh \
          --bam_fn=pbmm2/PlasmoDB-68_P_Genome/Cell1.bam \
          --ref_fn=ref/PlasmoDB-68_P_Genome.fasta \
          --threads=16 \
          --platform="hifi" \
          --model_path="${CONDA_PREFIX}/bin/models/hifi" \
          --include_all_ctgs \
          --output=`dirname clair3_Cell1/merge_output.vcf.gz` \
          --haploid_sensitive \
          --enable_long_indel \
          --no_phasing_for_fa