If so, by about how much? I want to know the downsides of skipping the phasing step on human microarray and WGS data.
Yes. Phasing data leads to a better imputation accuracy. More specifically:
- Improves allele matching: Correct phased data ensures that alleles are matched correctly between the reference panel and the target dataset.
- Using phased data, the computational time of imputation is significantly reduced.
- Having haplotypes (phased data) helps imputing variants that were not genotyped (missing variants/genotypes).
This paper based on the UK10K data has some actual numbers describing imputation accuracy of phased vs unphased data.