Interpretation of Genotypic association test

[Ana Ferreira]

Hello,

I'm studying a novel quantitative phenotype for which we know nothing about its genetic basis or inheritance pattern. I've already detected a few regions of interest that could underly the genetic basis of this phenotype, and now I'm testing the association of certain SNPs of interest under different models (additive, dominant, recessive, genotypic) to try to make an argument about the inheritance pattern of the associated alleles (are they dominant? recessive? additive? etc). I'm using PLINK1.9 with --linear under the genotypic model and I'm finding a hard time interpreting the output even after doing some extensive research over manuals and literature. I understand that the ADD term represents a test of additive effect (more copies of the allele increase value of the dependence). DOMDEV will test for a deviation for additivity (dominance deviation), and GENO_2DF with test both ADD + DOMDEV. 

This is the output for one of my SNPs. Both the ADD model and GENO_2DF are significant, but the DOMDEV model is not. Should I interpret this as there is no significant deviation from additivity (DOMDEV is non significant) so my heterozygote genotype is equally distanced from the reciprocal homozygote classes? In this case, how should I interpret a significant GENO_2DF model? Does it just reflect the fact that ADD is significant as well, but there shouldn't be a significant dominant deviation? Also, is it normal that GENO_2DF has no BETA value? 

CHR             SNP         BP   A1       TEST    NMISS       BETA         STAT            P 

   2   chr2_36198394   36198394    T        ADD       58      3.158        5.891    2.413e-07

   2   chr2_36198394   36198394    T     DOMDEV       58   -0.07089     -0.07867       0.9376

   2   chr2_36198394   36198394    T   GENO_2DF       58         NA        35.94    1.571e-08

Thank you in advance,

Ana F.

[Christopher Chang]

"Should I interpret this as there is no significant deviation from additivity (DOMDEV is non significant) so my heterozygote genotype is equally distanced from the reciprocal homozygote classes?"  Yup.

"In this case, how should I interpret a significant GENO_2DF model? Does it just reflect the fact that ADD is significant as well, but there shouldn't be a significant dominant deviation?"  That only tells you that ADD and DOMDEV are significant together.  It can be due to essentially just ADD (as in this case), just DOMDEV, or a combination.

"Also, is it normal that GENO_2DF has no BETA value?"  Yes, the underlying statistical test (approximate F-test) is different than the one used for the individual covariates.

[severa21]

 Hi ,

I have a question related to this post,

if the domdev simply indicates a deviation from additivity, how should I interpret the beta? I have noticed that often the sign is opposite to the beta of the additive model (see the attached file), is there a specific reason for it?
what is the most correct way to report the results of this analysis for a publication?  list all three tests (Add, domdev and Geno2f) for each SNP?
 
Thank you in advance