Attachment | Size |
---|---|
MVModel_ACE_OpenMx_ZN.R [6] | 44.23 KB |
Hi,
I am currently working on a multivariate correlated factors twin model with combined binary and continuous traits. I have six variables in total, the first two being binary data and the remaining four being continuous. I have been struggling for a while now in extracting meaningful output, specifically for the aetiological correlations. I receive genetic correlations with corresponding confidence interval lower bounds that are higher than the estimate itself (even though status is ok and model has converged).
For example:
confidence intervals:
lbound estimate ubound note
MZ.Acor[3,2] 0.4411063 0.4411063 0.999999 !!!
CI details:
parameter side value fit diagnostic statusCode method a11
1 MZ.Acor[3,2] lower 0.9285487 11548.08 success OK neale-miller-1997 0.6466693
2 MZ.Acor[3,2] upper 0.9999990 11548.08 success OK neale-miller-1997 0.6227424
This is the output to suggest the model has converged:
> AceFit$output$status$code
[1] 0
In case helpful, I am using the SLSQP optimiser and mxversion is 2.20.6
I have searched the openmx forums for different ways to adjust my script and troubleshoot (including adjusting 'function precision', mxTryHard/MxTryHardOrdinal with extratries) but have had no luck.
I am attaching my script here, though I cannot share the data. I appreciate this is quite a big model, so any tips and keen eyes on my script will be hugely appreciated!
Thanks in advance,
Zeynep