Create analysis module/notebook

[zhang-ivy]

It would be useful to have an analysis module that generates useful plots for diagnosing "simulation health", like the one in yank .

It might helpful to include some of the analyses used in our manuscript on investigating sampling issues in barnase:barstar:

• Free energy vs time plots (Figure 3C, D, E, F) -- including slope computation
  • scripts:
    • Generate a pickle file of the free energies at each time point: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/analyze_dg_timeseries.py
    • Generate plots: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/1_free_energy_timeseries_plots.ipynb
• dU/dlambda vs degree of freedom plots (Figure 4B, C, E, F)
  • scripts:
    • Generate a numpy array of dU/dlambda at each time point for each replica: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/run_du_dlambda_analysis_per_replica.py
    • Generate a dcd trajectories for each replica: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/run_make_traj_per_replica.py
    • Get dihedral angles for interface residues: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/get_residue_dihedrals_per_replica.py
    • Get distances between interface contacts: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/get_residue_distances_per_replica.py
    • Get water counts around mutating residue: https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/get_water_counts_per_replica.py
    • Plot dU/dlambda vs degree of freedom (aka feature): https://github.com/choderalab/perses-barnase-barstar-paper/blob/main/scripts/05_analyze/8_correlate_du_dlambda_with_features_per_replica_50ns.ipynb

[richardjgowers]

Just dropping some links for what openfe has done:

• Strutural analysis. This is using MDAnalysis, but hopefully this is getting relicensed soon https://github.com/OpenFreeEnergy/openfe_analysis
• Our energetics is mostly just a wrapper around MultiStateSamplerAnalyser: https://github.com/OpenFreeEnergy/openfe/blob/main/openfe/protocols/openmm_utils/multistate_analysis.py
• Plotting of all the above: https://github.com/OpenFreeEnergy/openfe/blob/main/openfe/analysis/plotting.py

In general, rather than writing structural analysis shouldn't be something FE specific, but instead you plug in the trajectory to mdanaylsis/mdtraj and can somewhat forget that it's FE-related except for the alchemical regions. We're currently tagging sections via "bfactor", to label them as core/A/B/environment etc.