search

HKU-BAL / Clair3

Clair3 - Symphonizing pileup and full-alignment for high-performance long-read variant calling
247 stars 27 forks source link

pileup variant for identical genomes #271

Closed lagphase closed 8 months ago

lagphase commented 9 months ago

Hi,

I have high quality ONT reads from a single cell that passed all the quality check. I assembled these reads into a genome, then used Clair3 on the reads and the assemble genome to see test out Clair3 accuracy. So theoretically I shouldn't see any SNP?

My command was in addition to the required flag --include_all_ctgs --no_phasing_for_fa --haploid_precise --call_snp_only

The output is:

[WARNING] No variant found, output empty vcf file [WARNING] Copying pileup.vcf.gz to /home/vdpham/Documents/dorado_basecalling/augWGS_2023/barcode11/porechop-flye-medaka-wf-alignment-clair3/clair3_haploidprecise/merge_output.vcf.gz

[INFO] Finish calling, output file: /home/vdpham/Documents/dorado_basecalling/augWGS_2023/barcode11/porechop-flye-medaka-wf-alignment-clair3/clair3_haploidprecise/merge_output.vcf.gz

So it's good that there was no variant found? But when I opened the merge_output.vcf file, it looks like this:

CHROM POS ID REF ALT QUAL FILTER INFO FORMAT SAMPLE

contig_1 58 . G . 11.56 RefCall P GT:GQ:DP:AD:AF 0/0:11:62:42:0.6774 contig_1 239 . C . 29.69 RefCall P GT:GQ:DP:AD:AF 0/0:29:63:54:0.8571 contig_1 966 . C . 21.30 RefCall P GT:GQ:DP:AD:AF 0/0:21:71:48:0.6761 contig_1 969 . C . 7.85 RefCall P GT:GQ:DP:AD:AF 0/0:7:71:39:0.5493 contig_1 970 . T . 26.99 RefCall P GT:GQ:DP:AD:AF 0/0:26:71:61:0.8592 contig_1 1211 . C . 18.24 RefCall P GT:GQ:DP:AD:AF 0/0:18:73:50:0.6849 contig_1 1214 . C . 16.75 RefCall P GT:GQ:DP:AD:AF 0/0:16:73:42:0.5753 contig_1 1215 . A . 23.37 RefCall P GT:GQ:DP:AD:AF 0/0:23:73:58:0.7945 contig_1 1219 . C . 17.63 RefCall P GT:GQ:DP:AD:AF 0/0:17:73:43:0.5890 contig_1 1222 . T . 23.90 RefCall P GT:GQ:DP:AD:AF 0/0:23:73:59:0.8082 contig_1 1227 . C . 24.19 RefCall P GT:GQ:DP:AD:AF 0/0:24:74:59:0.7973 contig_1 1231 . G . 23.76 RefCall P GT:GQ:DP:AD:AF 0/0:23:74:40:0.5405

How should I interpret this result? No variant found but there are pileup variants? What are pileup variants and are they true variants?

Thank you.

aquaskyline commented 9 months ago

These are RefCall, meaning that they are not variants. You can simply ignore the RefCall records.

On the other hand, since you have not enabled showing RefCall, but Clair3 is showing them, it means that Clair3 has a bug and we identified it already. The bug doesn't affect the correctness of the called variants, but was causing RefCall to be shown in the final VCF when no variant was called in the full-alignment calling stage.

A fix is scheduled for v1.0.6. Before that, ignoring all lines with the RefCall tag solves your problem.

lagphase commented 9 months ago

Hi,

Thanks for your response. I looked at the merge_output.vcf file again and there are 45 variants that are PASS and not RefCall, see below for an example (an extended image of the one in the previous post).

Most of the PASS variant QUAL score is below 10 but some are between 11 - 18. Should I ignore them because of the low QUAL score?

contig=

CHROM POS ID REF ALT QUAL FILTER INFO FORMAT SAMPLE

contig_1 58 . G . 11.56 RefCall P GT:GQ:DP:AD:AF 0/0:11:62:42:0.6774 contig_1 239 . C . 29.69 RefCall P GT:GQ:DP:AD:AF 0/0:29:63:54:0.8571 contig_1 966 . C . 21.30 RefCall P GT:GQ:DP:AD:AF 0/0:21:71:48:0.6761 contig_1 969 . C . 7.85 RefCall P GT:GQ:DP:AD:AF 0/0:7:71:39:0.5493 contig_1 970 . T . 26.99 RefCall P GT:GQ:DP:AD:AF 0/0:26:71:61:0.8592 contig_1 1211 . C . 18.24 RefCall P GT:GQ:DP:AD:AF 0/0:18:73:50:0.6849 contig_1 1214 . C . 16.75 RefCall P GT:GQ:DP:AD:AF 0/0:16:73:42:0.5753 contig_1 1215 . A . 23.37 RefCall P GT:GQ:DP:AD:AF 0/0:23:73:58:0.7945 contig_1 1219 . C . 17.63 RefCall P GT:GQ:DP:AD:AF 0/0:17:73:43:0.5890 contig_1 1222 . T . 23.90 RefCall P GT:GQ:DP:AD:AF 0/0:23:73:59:0.8082 contig_1 1227 . C . 24.19 RefCall P GT:GQ:DP:AD:AF 0/0:24:74:59:0.7973 contig_1 1231 . G . 23.76 RefCall P GT:GQ:DP:AD:AF 0/0:23:74:40:0.5405 contig_1 1236 . T . 27.41 RefCall P GT:GQ:DP:AD:AF 0/0:27:74:55:0.7432 contig_1 1495 . GA G 6.93 PASS P GT:GQ:DP:AD:AF 0/1:6:78:33,27:0.3462 contig_1 1713 . C CT 6.61 PASS P GT:GQ:DP:AD:AF 0/1:6:81:53,14:0.1728 contig_1 2007 . C . 21.95 RefCall P GT:GQ:DP:AD:AF 0/0:21:87:57:0.6552 contig_1 2008 . T . 24.87 RefCall P GT:GQ:DP:AD:AF 0/0:24:87:74:0.8506 contig_1 2436 . G . 20.41 RefCall P GT:GQ:DP:AD:AF 0/0:20:91:68:0.7473

aquaskyline commented 9 months ago

I think you should ignore them since they are indels with lower quality.

aquaskyline commented 8 months ago

Fixed in v1.0.6

bethsampher commented 6 months ago

Hi @aquaskyline , I was still having problems with this and noticed the fix only applies when --pileup_only is specified (see diff) and not when no variants are found with the full-alignment model. Please could you fix this? Thank you, Beth

aquaskyline commented 6 months ago

@bethsampher Please send your log file and show some VCF records that can describe your problem.