PheWAS for variant with only heterozygous cases

[h-desai]

Hello!

Currently working on generating PheWAS for 4 variants. Patients with two of the variants are homozygous (2), heterozygous(1), or controls (0) and PheWAS generates stats and the plot.
However, I am running into problems generating the stats and plots for variants that patients are heterozygous (1) or control (0) only.
I don't run into any errors when using phewas() and addPhecodeInfo() but there are no statistics generated at all.
When generating the phewasManhatten(), I get the following error:

Error in phenotypePlot(d, suggestive.line = suggestive.line, significant.line = significant.line, :
Annotate.phenotype.description must contain columns phenotype and description, or be TRUE with provided d$description.
In addition: Warning message:
In max(d$value) : no non-missing arguments to max; returning -Inf

For the two variants that I had success with, I changed the genotypes to only 0s (controls) and 1s (cases) and ran into the above problems.
For testing purposes, I added a random patient with a homozygous genotype (2) to the file and PheWAS generated the stats and plots.

Does PheWAS not work for variants that only have heterozygous cases?

[RobertJCarroll]

What do the results look like?
Specifically, did you review the note field of the output to see if there were issues using the model?
What does the model output look like when you do a single logistic regression on your own?

I don't have much to go on here, but it sounds like if you have zero variant homozygotes, the variants may be so rare that the model cannot converge. You can also try phewas_ext() and use firth regression, though there's only so much one can do if the variants are very rare in your population.