Hi there!
I have just imputed some low-coverage (<4x) whole genome sequencing samples using a WGS reference panel (>10x) and Beagle v5.4:
java -jar beagle.22Jul22.46e.jar gt=${file} impute=true gp=true ne=100 nthreads=16 ref=${refvcf} out=${outfile}
However, the resulting file only has "." in the INFO column so I am not sure how to assess quality. Can I please have some advice on why Beagle is not giving me any accuracy scores or info like Allele Frequencies or Dosage R-squared in the output? I have no idea why this is happening, and I need them to assess accuracy!
From the manual it states:
The vcf.gz file is a bgzip-compressed VCF file that contains phased, non-missing genotypes for all non-reference samples. The output vcf.gz file can be uncompressed with the unix gunzip utility. If a reference panel is specified and ungenotyped markers are imputed, the VCF INFO field will contain: • A “DR2” subfield with the estimated squared correlation between the estimated allele dose and the true allele dose • An “AF” subfield with the estimated alternate allele frequencies in the target samples • The “IMP” flag if the marker is imputed
Is it possible that this is because I am working with WGS data, not SNPchip data, so I am only imputing based on known SNPs/genotypes within the reference panel, so no new SNPs are imputed? And AF and DR2 are somehow related to that?