I really like datreant to add arbitrary meta-data to my simulations that
I can easily filter later. But datreant has one big drawback when I want
to recommend it to others. It requires that any analysis is done in
python. That is a shame since I think datreant can also help people who
don't use python for their analysis.
To fix this I created a new datreant project **datreant.cli**
This project installs a script **dtr** that can be used to filter and
add Treant objects on the command line. So to cluster several
simulations one could do
for f in $(dtr search simulations --tags=1AKE,charmm32)
gmx cluster -s system.gro -f simulation.xtc
The search command always returns a list of absolute paths.
I haven't tested this project yet just some short checks that it works.
It doesn't work with Sims yet since that requires me to depend on
mdsynthesis, I would like a general solution where that isn't the case.
Filtering by categories is also not possible currently. That is mainly
because I don't know how people would like this to behave and what a
good API is. I welcome any suggestions.
It should also help adoption of datreant when we can tell people that
they are still able to use their old tools but still enjoy the benefits
of having meta-data attached to simulations.
Note: For me the current state of this is enough. I will only add new
things sporadically but I welcome any PR or other help to improve this.