Closed tgong1 closed 1 year ago
No. We didn't test Illumina SNPs. It also depends on how you called your SNPs for Illumina and ONT. You may check the evaluation by the PrecisionFDA challenges. If Dragen was used for Illumina-SNP calling, the Illumina accuracy should be better than ONT. But if they were not called by Dragen, I am not sure which one would be better than the other. I suspected only marginal differences. So you may stay with the Illumina SNPs.
Thank you for the quick reply. I don't think the SNP we have called by Dragen. We are considering to have a quick comparison/benchmarking. Do you mind sharing the link of GIAB phased benchmark SNP and SV of HG002, HG003 and HG004 you have mentioned in your paper with me?
Thank you, Tingting
We're trying this out too using SNPs from bcftools. So far looks good.
Hi @tgong1 This is the GIAB benchmark mentioned in the paper.
URLs of GIAB v4.2.1 benchmark for HG002/HG003/HG004
HG002
https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG002_NA24385_son/NISTv4.2.1/GRCh37/HG002_GRCh37_1_22_v4.2.1_benchmark.vcf.gz
HG003
https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG003_NA24149_father/NISTv3.3.2/GRCh37/HG003_GRCh37_GIAB_highconf_CG-IllFB-IllGATKHC-Ion-10X_CHROM1-22_v.3.3.2_highconf.vcf.gz
HG004
https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG004_NA24143_mother/NISTv3.3.2/GRCh37/HG004_GRCh37_GIAB_highconf_CG-IllFB-IllGATKHC-Ion-10X_CHROM1-22_v.3.3.2_highconf.vcf.gz
URLs of CMRGv1.00 curated SVs https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG002_NA24385_son/CMRG_v1.00/GRCh37/SupplementaryFiles/HG002v11-align2-GRCh37/HG002v11-align2-GRCh37.dip.vcf.gz
Many thanks for the help!
Hi longphase team,
Thank you for this helpful software. I'm recently using longphase in our cohort.
We have SNP/indels called from Illumina short-read NGS data. So I have used Illumina SNP VCF + ONT SV VCF + ONT BAM as input for longphase and it has ran pretty well. However, I was wondering the accuracy (or switch rate) of SNP and SV phasing.
I saw you've included a comprehensive benchmarking on GIAB's data in your paper. Have you test the performance of longphase using SNP from short-read NGS data? Will using SNP called from ONT have better performance on phasing from your experience?
Thank you, Tingting