treangenlab
No improved phasing
Hi,
We are trying to phase a region, but there remains a gap which cannot be phased using SNPs. Now we have tried the high-accuracy and high-success parameters of MethPhaser, but still we are unable to close the unphased gap. For this assembly we have methylation-called HiFi reads (N50 = 14kb) and ONT reads (N50 = 45kb) combined.
We suspect that one haplotype probably has an insertion at this position on one haplotype, which is absent on the other haplotype. Are there other parameters in MethPhaser, for example lowering the required coverage, to phase this difficult region? Could lowering or increasing the -c and -a parameters help?
This is our command:
meth_phaser_parallel -b haplotagged_pri_sorted.bam -r collapsed_EAW_IGK.fa -g whatshap_phase.gtf -vc phased_whatshap_sorted.vcf.gz -o methphaser_output -ml -2
and
meth_phaser_post_processing -ib haplotagged_pri_sorted.bam -if methphaser_output/ -ov methphase_whatshap_out_hs.vcf -ob methphase_whatshap_hs -vc phased_whatshap_sorted.vcf.gz -hs
With or without the -hs flag.
Hopefully you can hulp us to solve this phasing problem.
I think yes lowering -c and -a could be helpful, but this is a 100kb gap, is this in centromere region? might be too long for MethPhaser. However, happy to talk more if you need help.
Lowering -c and -a did not affect the phasing unfortunately. Any other suggestions? It is not a centromere region, but a sequence more similar to the HLA region. So we are aware that phasing is difficult, but we hoped to solve it with this tool.
I just checked with my colleagues that are more familiar with HLA regions. We all found the coverage is not very nice in this region, and it would be better if you can group the reads to haplotypes on IGV so that it is better to see. But we agreed with you about the HLA region is hard to phase.. Sorry but I don't think MethPhaser could be super helpful in this context given the gap might be too large. Also you may try population phasing to stich this gap but that's mainly for human.. Let me know if you want to explore more, like phasing with or without MethPhaser
I indeed also have the image with the haplotypes colored. And it clearly shows a grey area in the gap location.
We can do population phasing as well, but we only have a parent/child couple, which means we lack a parent for trio-binning. Do you have any suggestion that would be able to phase using a duo?
My intuitive hunch of this is that you can use child sample twice as one of the parent so that you can do trio phasing. Or you can try https://odelaneau.github.io/shapeit5/