PLINK p value returning NA

Question

[Sin Yu, Lin]

Hi everyone,

I've been analyzing a biobank data this week for several times, if only doing --assoc test, the results are fine. But as I try to do --logistic or --linear, the p values or ORs are all NA. The original fam file doesn't include gender and phenotype data (which is all 0 and -9), so I write the phenotype data (both quantitative and qualitative) in phenotype.txt and gender data in covariate.txt. Other variables such as BMI, age and PC1 to PC10 are also included in the covariate.txt. I notice that the PLINK website suggest that there might be a strong multi-collinearity between my covariates causing the result, but I don't think that I have this problems (as you can see they are quite normal). I've also tried not to do --allow-no-sex or not putting sex into the --covar-name, but all the results remain NA, which is quite frustrating. I just started to learn PLINK in the last few weeks by myself, so maybe I'm missing something important, any help will be a lifesaver.

---

--bfile myfile --allow-no-sex --remove-fam remove.txt --geno 0.02 --mind 0.02 --maf 0.05 --hwe 1e-6 --pheno phenotype.txt --pheno-name pheno1_CLS --logistic --ci 0.95 --covar covariate.txt --covar-name SEX,AGE,BMI,PC1-PC10 --make-bed --out result

Answer 1

[chrchang523]

If you run this with plink 2.0 instead of plink 1.9, there will be an ERRCODE column in the output indicating the reason for each "NA" p-value (https://www.cog-genomics.org/plink/2.0/assoc#glm_errcode ).

[Sin Yu, Lin]

I'll try this right away, thank for your advise!

[Jingjingzhang]

I run this gwas pipeline with PLINK v2.00a2.3LM 64-bit Intel (24 Jan 2020). But I also met this question ( PLINK p value returning NA ).

[chrchang523]

Please update your plink 2.0 build.