Consider classes such as:
System / Lambda / Replica
Each replica should record its source. Should be serialisable in both directions with JSON. Should have an export function to pandas.
A user should be able to "add system" which will discover and assemble the information together.
The major aspect of this is the going to be the ability to use these classes in:
regardless of the number of replicas, lambdas, settings
unit testing etc
its use in the ability to do "online" simulations, ie adaptive lambdas, replica, length
The last means it will become a part of the "mpi manager" or something akin
Another level of analysis includes often the same protein and many ligands, therefore:
Metaanalysis:
Protein System:
System (Protein, Ligand)
Lambda
Replica
This way you could include in your metanalysis any set of studies, and compile them
Consider connecting the simulations data to other databases, particularly linking to the databases with the experimental errors. Ie if the binding energies have been uploaded for similar compounds, same compounds, etc, which can be used as a reference. But also, if the binding energies are uploaded - if there are errors attached to them.
Analysis module features:
function for checking the convergence of each replica, each energy component
Analysis of the actual data:
RMSD, binding pocket and its properties, across ligands etc,
protein oriented analysis
ie standard protein analysis,
how its behaviour is affected by different ligands,
Consider classes such as: System / Lambda / Replica
Each replica should record its source. Should be serialisable in both directions with JSON. Should have an export function to pandas.
A user should be able to "add system" which will discover and assemble the information together.
The major aspect of this is the going to be the ability to use these classes in:
The last means it will become a part of the "mpi manager" or something akin
Another level of analysis includes often the same protein and many ligands, therefore:
This way you could include in your metanalysis any set of studies, and compile them
Consider connecting the simulations data to other databases, particularly linking to the databases with the experimental errors. Ie if the binding energies have been uploaded for similar compounds, same compounds, etc, which can be used as a reference. But also, if the binding energies are uploaded - if there are errors attached to them.
Analysis module features:
Analysis of the actual data: