Notebook + sample data for estimating FE and uncertainty

[melissapathil]

/playground/notebooks/pale_toolkit.ipynb initial notebook for ddG and uncertainty BAR estimates with NEQ cycles
/playground/protein-mutation/ABL-ATP directory with sample data for Abl-ATP

Estimating $\Delta\Delta G$ and Uncertainty using PyMBAR

• is_cycle_complete : check if given CycleUnit directory has completed results
• get_num_cycles : get minimum number of completed cycles across mutations to compare BAR estimates
• load_work_arrays : load work values stored in npy arrays of completed cycles
• get_uncertainty: estimate uncertainty by taking standard deviation of bootstrapped BAR-estimated free energies
• plot_works : plot work trajectories and distributions
• compute_ddG_estimate : compute ddG and uncertainty using MBAR

[ijpulidos]

At first glance, I noticed that we probably don't want to add all the serialized JSON files in the code base. I presume they should be easy to generate with the notebook, correct? If not, then we just probably want one or two examples that can be used to illustrate the functionalities implemented in the notebook.

EDIT: Similarly with the PDBs

[ijpulidos]

These json files from the ABL+imatinib notebook, they were probably added to the commit by mistake.

[melissapathil]

I added the notebook for FE estimates and plotting work trajectories/distributions. I'm having trouble updating the branch with scripts and data for Abl-ATP but I'll try to get that up soon.

Update: scripts, data, and analysis notebook added! I'll convert this to a PR after adding some notes.

[ijpulidos]

Great work! The notebook is looking great.

I made some comments, but overall just avoiding having too many files (maybe we want to showcase the toolkit/utilities just with a couple of cycle results or so)

[ijpulidos]

We don't want to be storing the checkpoint files in the repo.

[ijpulidos]

The notebook looks very good, I think we probably want to have one that showcases the functions (shows plots, ddg estimates, etc.) for a couple of cycles or something like that if possible. That way we don't have to store all the results files which are not great to have on the repo.

We probably also want the notebook having the outputs (at least the "useful" ones, like plots and DDGs or similar).

[ijpulidos]

Same here, we don't want to store checkpoints.

[ijpulidos]

Samples of these plots should probably be in the notebook itself. That way we can showcase the tools and the results in the same file, and avoid having PNG(binary) files around as well.

[ijpulidos]

I don't think we need to store these scripts to run the DAGs as part of this set of changes. If there are improvements from the current scripts in the playground directory we probably want to make a new PR improving these, or adding new ones if needed.