Please post comments & responses here regarding the Twin Workshop 2010 (schedule, ideas on what to teach, ideas on how to teach, random thoughts, etc.).
I wanted to check the official party on commands to save the output from openMx - this is something that I will need to address on the tues morning
Are we suggesting people save the R-workspace or are you going to introduce some of the "save to text" functions?
I'm particularly thinking of the situation in which R crashes or your unix job hits a wall time before the script completes.
With Mx1.0 output wrote at checkpoints so scripts that crashed always left a 'paper trail' but I haven't been able to get this running smoothly under openMx
Checkpoints (and restoring from checkpoints) are not yet implemented in OpenMx, but are at the top of the list of features to add in the near future.
Just anf FYI in case anyone stumbles on this thread: checkpointing has been implemented as an option for mxRun()
mxRun(..., checkpoint = TRUE)
You can set things like checkpoints and parallel execution on multiple cores as global options like this:
# set up number of cores
mxOption(NULL, "Number of Threads", omxDetectCores() - 1)
mxOption(NULL, "Always Checkpoint", "Yes")
mxOption(NULL, "Checkpoint Count" , 1)
mxOption(NULL, "Checkpoint Units" , "evaluations")
The main web page for the 2010 workshop is at http://ibg.colorado.edu/dokuwiki/doku.php?id=workshop:2010:start
The schedule is at http://ibg.colorado.edu/dokuwiki/doku.php?id=workshop:2010:schedule
Most workshop faculty will have logins for the wiki, but if not (or if you don't remember your password) go to https://ibg.colorado.edu/dokuwiki/doku.php?id=workshop:2010:schedule&do=... and register or request a new password.
I wonder if this particular discussion group would be better labeled Twin Workshop 2010?
<5 mins later>
Ok... hardly any sooner noted than done!
I'll start it off with an email I wrote Mike regarding topics at the workshop. He suggested I post here:
Given that the workshop participants will now be working in R, I wonder if there should be a session on simulating data from the ground up (e.g., not using a particular program/script, but rather how to simulate twin/family data using base R). Simulating data in R (which is quite easy) and then running it in OpenMx is good not only for checking that models are working as expected, but also for helping to reinforce conceptual understanding of these models. Thoughts?
Simulation is a good suggestion. We figured that it would be covered in
the power & sample size session on Thursday afternoon. However, it perhaps needs more attention than that because it is such a fundamental technique for establishing the statistical validity of anything. Mainly because we're not as smart as Fisher, so showing things from first principles is less of an option :). Putting a smattering of mvrnorm() into the Tuesday morning R session might be a good place to start.
I am a computational biologist very new to Twin analyses. I wanted to get some suggestions from you regarding the Twin analysis. In my data set I have 24 MZ twins with a mix of some same sex male twins and some female and their age ranges from 21- 90 yrs old. Similarly, I have 10 DZ pairs with similar combination of male and female twin pairs and age range. How do I adjust the age and gender effect which might have effect on the variance estimate.
Any input would be helpful