Towards accurate, force field independent conformational ensembles of intrinsically disordered proteins
This is the github repository that includes all the notebooks and scripts to reweight MD simulations of intrinsically disordered proteins using a multiple experimental data-types. The notebooks are all arranged in their own separate directories. The scripts to calculate the predicted data from forward models such as Pepsi-SAXS, SPARTA+ and PALES are also included herein.
To learn more about our Maximum Entropy Reweighting Method please check out our paper/preprint:
-
Download the Max_ent_env.yml file and activate the environment by "conda env create -f Max_ent_env.yml"
-
Run the calc_exp.py script as follows:
- ./calc_exp.py <xtc/dcd>
- You can add flags for whatever experimental data you need calculated eg --cs --rdc --jcoupling --pre
- Add the residue numbers for which you want to calculate the PRES for (eg for ACTR ./calc_exp.py --pre 3 21 41 61)
- The script requires SPARTA+ and PALES installed in your environment
- Run the Pepsi-SAXS calculation script as follows:
-
-./calc_saxs.py <pdb_dir> <exp_file> <work_dir> --nproc --nchunk (the script requires the user to have the Pepsi-SAXS executable in the working directory or for the user to point the script towards the executable by editing the script. The script uses multiprocessing, and you can customize the number of processors and chunk size you want to use, with --nproc for number of processors and --nchunk for chunksize on each processor.
-
If Pepsi-SAXS is not automatically installed when activating the environment, it can be downloaded from https://team.inria.fr/nano-d/software/pepsi-saxs/
-
After all the calculations are done have the directory set up as follows: https://drive.google.com/drive/folders/1wROMvOgKjLkhjBD4FRiT0OdWTZyDupgM?usp=sharing. The drive link has all the experimental and ensemble computed data for the a99SBdisp, c22star and Charmm36m trajectories for PaaA2, ACTR, drkN, alpha-synuclein and Ab40 proteins.
-
For any protein system run the {protein}_{forcefieldname}_reweight.ipynb notebook. This notebook does the actual reweighting. After all the rounds of reweighting have been completed, the notebook automatically saves all the relevant information and weights in their own separate directories.
-
After the reweighting notebook, for the same protein system run the {protein}_{forcefield}_analysis.ipynb notebook This notebook loads the relevant dictionaries that were generated in step 5, and plots all the necessary tables and figures.
-
Run the {protein}_{forcefield}_ensembles.ipynb notebook. Requirements for the notebook are:
- a structure file such as a pdb file for your protein system
- an xtc or dcd file for your simulation
- a perfectly ideal helix structure for alpha-helical order paramter calculations (can be generated by the make_helix.py script)
- weights from your reweighting run (We use the weights from the combined data-type as restraint run, but any weights would do) This notebook loads the relevant required files and calculates the before and after reweighting ensemble properties such as secondary structure propensities, inter-residue contact maps, Radius of gyration distributions etc.
If you have any queries about the paper and/or the code please reach out to: kaushik.borthakur.gr@dartmouth.edu, paul.j.robustelli@dartmouth.edu