Lost haplotype after hapcorrection

[pabloangulo7]

Hello,

I successfully ran the hapcorrect module; however, when I looked at the results I had a question about the phasing correction. In the images I show a chromosome before and after the correction. Without the correction you can see that the coverage for both H1 and H2 is in similar ranges at all points. However, after the correction only the block for Haplotype 1 is visible, indicating that the entirety of the allele frequency and depth corresponds to Haplotype 1, while Haplotype 2 disappears. Therefore, I’m not sure whether this is an expected result or if there is some kind of error.

Thanks in advanced,

Pablo

[tahashmi]

Hi @pabloangulo7 , this looks like LOH region. Could you check snps_loh_plots dir? It should have a output
<genome-name>_genome_loh.html which shows genome-wide LOH plot (in tumor-only mode). Also no need to use smoothing option.

[pabloangulo7]

Yes, this plot would be the snps_loh_plot for the same chromosome:

I notice that in the original phased vcf file there are many SNPs where the GT is 1/1 and the AF value is between 0.9 an 1, but then in the rephased.vcf file, the GT changes to 1|0. I don't know if that could be one of the reasons.

Eg:

Original
chr15 53078752 chr15_53078752 G A 33.15 PASS P GT:GQ:DP:AD:AF:PS 1/1:33:65:0,63:0.9692:.

Rephased
chr15 53078752 chr15_53078752 G A 33.15 PASS P GT:GQ:DP:AD:AF:PS 1|0:33:65:0,63:0.9692:24800001

The command I used is:

python ${wakhan} hapcorrect --threads ${threads} --reference ${ref_genome_fasta} --target-bam ${bam_file} --tumor-phased-vcf ${snp_phased_vcf} --genome-name ${sample} --out-dir-plots ${HAP_correction} --loh-enable --phaseblocks-enable

Could you help me to understand the output? Would you still recommend me to use --loh-enable option? Thank you very much in advance,
Pablo

[tahashmi]

So basically, above plot is based on Hets/Homo SNPs ratios in bins. So it looks like Homo SNPs have higher ratio and LOH is being detected, so this is real LOH. Then during rephasing, Wakhan assign that LOH region to a single HP, that's why you get that altered SNP position with phased GT field. --loh-enable is only to show LOH region nothing to do with functionality. In tumor-only mode, Wakhan detects LOH regions by default.