Implement an adaptive search for MD simulations run with AMBER.
Use your own msmbuilder defined models for the search.
Clone and install from source
git clone https://github.com/jeiros/msmadapter.git
cd msmadapter
pip install -e .A crude example of the necessary directory structure to get started:
.
├── [ 50] generators
│ ├── [ 66] gen1
│ │ ├── [ 575] Production.in
│ │ ├── [ 33] seed.ncrst
│ │ └── [ 36] structure.prmtop
│ ├── [ 66] gen2
│ │ ├── [ 575] Production.in
│ │ ├── [ 33] seed.ncrst
│ │ └── [ 36] structure.prmtop
│ ├── [ 66] gen3
│ │ ├── [ 575] Production.in
│ │ ├── [ 33] seed.ncrst
│ │ └── [ 36] structure.prmtop
│ └── [ 66] gen4
│ ├── [ 575] Production.in
│ ├── [ 33] seed.ncrst
│ └── [ 36] structure.prmtop
└── [ 523] msmadapt.py
Where msmadapt.py is the script that controls the adaptive search logic.
A lot of it can be left as defaults:
from msmadapter.adaptive import App, Adaptive
app = App(from_solvated=True)
ad = Adaptive(app=app, stride=1, atoms_to_load='not water',
sleeptime=6*3600)
ad.run()Then start the adaptive sampling scheme with the following command:
nohup python msmadapt.pyThe nohup is to be able to detach from the computer where you're running this.
You can trace the output of the search on the contents of the nohup.out file that will be generated.