warning: Skipping --glm regression on phenotype 'standing_height_f50_0_0

[ארז דור‎]

Hi everybody,

I have a question and I will describe the stages I did so far:

1) first I did QC with some paramters. I also removed samples by ID which have high relatedness

2)In order to get plink to see the assesment_center as catgorial variable, I added 'C' before each  assesment_center number

"awk 'NR>1 { $46="C"$46 }1' covariates_age2.txt > covariance_matrix"

3) There were two "sex" variables, so I deleted one col (sex_f31_0_0)

"cut -d' ' -f1-44,46- covariance_matrix > covariance_matrix 2" 

4) I normilaized the covariance matrix

"plink2 --bed GEN/ukb_cal_chr1_v2.bed --fam GEN/ukb56774_cal_chr1_v2_s488264.fam --bim GEN/ukb_snp_chr1_v2.bim --covar phenotypes/covariance_matrix2 --covar-variance-standardize --write-covar --out phenotypes/covariance_matrix_standardize  "

5) I splitted to catagorial variables

"plink2 --bed GEN/ukb_cal_chr1_v2.bed --fam GEN/ukb56774_cal_chr1_v2_s488264.fam --bim GEN/ukb_snp_chr1_v2.bim --covar phenotypes/covariance_matrix_standardize.cov --split-cat-pheno omit-most covar-01 genetic_sex_f22001_0_0 uk_biobank_assessment_centre_f54_0_0 --write-covar --out phenotypes/covariance_matrix_split_dummies_standardize"  

6)Than I tried to do linear regression to standing_height_pheno (the command is found at the top of the screen photo)

but I am getting a warning (as seen in the photo) and its not working

Can anybody help me please?

Thank you

[Christopher Chang]

"You may want to remove redundant covariates and try again."  Are all 64 covariates (actually, more than that if you have a categorical covariate in the mix) really worth including in this regression?

After you've given at least a little bit of thought to that question, see the section of https://www.cog-genomics.org/plink/2.0/assoc on multicollinearity.

[ארז דור‎]

?So, which covariates should I remove in order to do a linear regression without  multicollinearity  

ב-יום שבת, 28 בנובמבר 2020 בשעה 19:43:20 UTC+2, chrch...@gmail.com כתב/ה:

[Sam Choi]

Some good starting points are:

https://datascienceplus.com/multicollinearity-in-r/

https://campus.datacamp.com/courses/human-resources-analytics-predicting-employee-churn-in-r/predicting-turnover?ex=9