Richard

its mainly the lack of graphics on clusters, but exacerbated by the fact that I want to generate screenshots for around 50k datasets. Is the plotting option based on tkmedit...

we are thinking of using something akin to braindr to do a fact visual qc on output images. so far I've not een able to find a tkmedit independent solution...

Apologies for totally missing this thread... In the end we ended up creating screenshots with freeview by simulating a graphics server that runs on our cluster: Something like: ``` xvfb-run...

This is the rough flow ``` cd ${workdir} echo freeview -v ${subjectdir}/mri/brain.mgz -f ${subjectdir}/surf/lh.pial:edgecolor=red:edgethickness=1 -f ${subjectdir}/surf/rh.pial:edgecolor=red:edgethickness=1 -f ${subjectdir}/surf/lh.white:edgecolor=blue:edgethickness=1 -f ${subjectdir}/surf/rh.white:edgecolor=blue:edgethickness=1 > freeviewconfig_multi.txt echo -viewport coronal -slice 102 128 100 -ss...

At least this has proven to be SLURMable for large dataset on a HPC where we can't open graphical windows.


I would normally agree, but we are generating about 10-15 images per subject and doing this on a 150.000 subject dataset, I think having more or fine-grained images is just...

This might be helpful: https://github.com/rb643/imaging_transcriptomics/blob/master/gene_decode.R Its essentially a wrapper around the Enrichr package in R and could potentially be stripped down to be a bit more general