We use snptest to test associations between disease status and genotypes. We see that the BETA and the OR in our results do not match.
In the following example, frequentist_add_beta_1=0.35829 so we expect OR=exp(frequentist_add_beta_1)=1.430881. But the result file lists all_OR=1.47457. The difference is big enough that it is not caused by rounding ( exp(0.358285)=1.430873; exp(0.358295)=1.430888)
Can somebody shed light on this confusion we have? Thank you!
Shengchao
# Analysis: "SNPTEST analysis, started 2020-04-16 02:16:37"
# started: 2020-04-16 02:16:45
#
# Analysis properties:
# -chunk 1 (user-supplied)
# -data 1.110229787_cond.vcf.gz pheno_snptest1.def (user-supplied)
# -frequentist 1 (user-supplied)
# -genotype_field GP (user-supplied)
# -method score (user-supplied)
# -o condition/1.110229787.out (user-supplied)
# -pheno case_control (user-supplied)
#
alternate_ids rsid chromosome position alleleA alleleB index average_maximum_posterior_call info cohort_1_AA cohort_1_AB cohort_1_BB cohort_1_NULL all_AA all_AB all_BB all_NULL all_total cases_AA cases_AB cases_BB cases_NULL cases_total controls_AA controls_AB controls_BB controls_NULL controls_total all_maf cases_maf controls_maf missing_data_proportion het_OR het_OR_lower het_OR_upper hom_OR hom_OR_lower hom_OR_upper all_OR all_OR_lower all_OR_upper frequentist_add_pvalue frequentist_add_info frequentist_add_beta_1 frequentist_add_se_1 comment
...other rows snipped...
1:110229765 1:110229765 01 110229765 A C 6151 1 1 170 745 1230 0 170 745 1230 0 2145 71 337 694 0 1102 99 408 536 0 1043 0.252914 0.217332 0.290508 0 1.15172 0.822031 1.61363 1.80539 1.30447 2.49867 1.47457 1.28385 1.69361 1.1534e-07 1 0.35829 0.0675925 NA
...other rows snipped...
I'm pretty sure that the calculation for both differs. But unfortunatly I can't figure it out yet.