I am new to GCTA and was trying to use GSMR to investigate the association between blood pressure (UKBB+ICBP) and breast cancer (BCAC),but I got this error:
GSMR analysis: 1361 instruments loaded.
There are 1361 SNPs in the dataset.
918 SNPs were retained after filtering.
848 SNPs were retained after the HEIDI-outlier analysis.
Estimating bxy using all the remaining instruments.
Error in bxy_gsmr(bzx, bzx_se, bzy, bzy_se, ldrho) :
The variance-covariance matrix for bxy is not invertible!
In addition: Warning messages:
1: In filter_summdat(snpid, bzx, bzx_se, bzx_pval, bzy, bzy_se, bzy_pval, :
224 non-significant SNPs were removed.
2: In filter_summdat(snpid, bzx, bzx_se, bzx_pval, bzy, bzy_se, bzy_pval, :
There were SNPs in high LD. After LD pruning with a LD r2 threshold of 0.05, 219 SNPs were removed. Note: The threshold of LD can be changed by the "ld_r2_thresh".
Can you please tell me what does the error (The variance-covariance matrix for bxy is not invertible) mean and how to solve it ? I have used Plink to remove the SNPs in LD (r2 < 0.05) before running the analysis.