Error message run in SSH for day 7 stretch goal 1 GCTA practice

ISG Workshop 2022 Day 7
1

Running below coding in SSH for day 7 stretch goal 1 GCTA practice, I got an error message.

gcta64 --pheno ozht.phen --qcovar ozht.prs1.qcovar --covar ozht.covar --mgrm GRM-CRM.txt --out ozhtPRS1.ACE --reml --reml-est-fix.

********* Below is the part of the error message*********
Performing REML analysis … (Note: may take hours depending on sample size).
1894 observations, 8 fixed effect(s), and 3 variance component(s)(including residual variance).
Calculating prior values of variance components by EM-REML …
Updated prior values: 29.7308 29.7308 24.092
logL: -4731.45
Running AI-REML algorithm .
Iter. logL V(G1) V(G2) V(e)
1 -4688.37 30.35419 41.91104 12.26942
2 -4597.70 32.14758 44.64333 8.91853
3 -4567.86 31.04674 49.28839 7.57450
Error: the information matrix is not invertible.
An error occurs, please check the options or data
*********** End of the error message*********

How to fix this problem?

2

hmm that is unexpected.

Try going back to the step where you make the CRM and rerun from there. I wonder if there is an issue with that.

3

Yes, this gcta error often relates to colinearity between the random effects or between the covariates. It can also happen when covariates that are constant across all participants, or if you include the outcome variable as a covariate, by mistake (this has happened to me before :sweat_smile:)