how to define minMAC for generic burden test?

[sinhanushka12]

I'm trying to run the code below, and for each gene I want to generate the minimum number of masks possible. I’m using only variants with MAF < 10⁻³, and my mask definition contains just one mask. However, with the settings below, I end up with two masks total, and most genes fail the MAC filter even though I have set --minMAC 0.5 (the lowest possible value).

regenie \
        --step 2 \
        --bgen ukb23159_c${chr}_b0_v1.bgen \
        --sample ukb23159_c${chr}_b0_v1.sample \
        --keep ${FAM_FILE} \
        --phenoFile ${PHENO_FILE} \
        --covarFile ${COVAR_FILE} \
        --phenoCol ${PHENO_COL} \
        --anno-file anno_maf_1e-3_chr${chr}.tsv \
        --set-list gene_set_maf_1e-3_chr${chr}.tsv \
        --mask-def ${MASK_DEF} \
        --aaf-bins 0.01 \
        --write-mask \
        --write-mask-snplist \
        --pred ${PRED_LIST} \
        --bsize 400 \
        --bt \
        --minMAC 0.5 \
        --firth --approx \
        --rgc-gene-p \
        --gz \
        --verbose \
        --threads 2 \
        --out new_regenie_step2_imbalanced_binary_50_causal_simulation_chr${chr}

It seems unlikely that so many genes would fail both masks because of MAC < 0.5. Could you please advise on what might be going wrong or what I should look into?
I get the following error for most of the genes I'm testing:

set [1/517] : ENSG00000000419 - 195 variants...1 chunks
     -reading in genotypes, computing gene-based tests and building masks...WARNING: 2/2 masks fail MAC filter and will be skipped...done (835ms) 
 set [2/517] : ENSG00000019186 - 459 variants...2 chunks
     -reading in genotypes, computing gene-based tests and building masks...WARNING: 2/2 masks fail MAC filter and will be skipped...done (1979ms) 

[joellembatchou]

Hi,

Please include --check-burden-files to see if there is any discrepancy between the input annotation files. You can also check the output mask snplist files to see how many variants are included in the masks. Also with BGEN input files, check the format for the allele order; if the first allele is REF, then you should be using --ref-first (REGENIE assumes the REF allele is last).

Cheers,
Joelle