Phasing with Beagle 5.2 and no reference panel

Question

[Hana]

Hi everyone,

I have a question about phasing with Beagle 5.2 without a reference panel. I have seen in answers in a couple other posts about Beagle that trying to phase with too few samples and no reference panel is not advised. I was wondering if anyone had any insights into what constitutes "too few samples"? Or any suggestions where I could go to learn more about this? I have a multi-sample vcf file with 24 samples that I generated with GATK HaplotypeCaller that I was hoping to phase, but I do not have a reference panel. After looking into this I am now wondering if I have too few samples for this analysis.

I would really appreciate any insights you may have! Thank you

Answer 1

[4galaxy77]

24 is too few in my opinion, although it's just an opinion rather than backed up by any hard facts. It probably depends on some factors like population size / species / how inbred/outbred the organism is.

I would reccomend either a) downloading and using a reference panel such as the 1000 genomes or b) use a service such as the Sanger Imputation Service or TopMed, which also use a reference panel (but you don't have to download it).

[Hana]

thank you! I will look into those options