how to deal with "The number of covariates exceeds the number of samples" on matrixeqtl

Question

[Binghong]

Hi guys, I try to analyze the association of SNP and gene using the matrieqtl, but there are 9094 covariates but only 171 samples, the matrixeqtl show that "The number of covariates exceeds the number of samples." I have no idea how to deal with it, how to reduce the number of covariates. Any guidance is appreciated !!

Answer 1

[LChart]

This just means your metadata has too many rows. Usually this file will contain biological age, sex, maybe 1-3 genotype PCs, and maybe 1-10 expression factors (PEER or SVA or something); and not much else. What do you have in yours?

[Binghong]

I added the CNV data to the covariates, which increased number of covariates to 9,000

[LChart]

You can only retain a small number of CNVs; or you can compute a SVD of the CNV genotype matrix and include the top PCs

[Binghong]

Thanks for your guidance. I will have a try