This code is pretty much an automatization tool that follows the DADApy official tutorial, but it makes it easier for AMBER users because AMBER netcdf files can be provided as input directly.

It can extract temporal courses of dihedral angles from your trajectory file and analize them with the DADApy library using the Advanced Density Peaks (ADP) algorithm. It recieves a file with such time courses instead if you prefer so.

It is recommended to check the DadaPy and ADP papers:

• Glielmo, A., Macocco, I., ... & Laio, A. (2022). DADApy: Distance-based analysis of data-manifolds in Python. Patterns, 3(10).

• Rodriguez, A., & Laio, A. (2014). Clustering by fast search and find of density peaks. Science, 344(6191), 1492-1496.

Also, make sure to check the DADApy library repo:

https://github.com/sissa-data-science/DADApy

• Python 3
• Libraries: DADApy, NumPy, ParmED (for netcdf), ASE (for xyz files), matplotlib.

Run the script from the command line with the required arguments.

python D-clust.py [-h] -i INPUT -f FORMAT [-d DIHELIST] [other optional arguments]

• -i, --input: Input file name (required).
• -d, --dihelist: A text file with the atom index of each dihedral to be extracted (required unless the format is 'dihe').
• -f, --format: Input file format ('xyz', 'netcdf' or 'dihe', required).
• -id, --id: Intrinsic dimension (integer, if not provided, it will be estimated).
• -v, --visualize: Plot on screen the intrinsic dimension scaling ('True' or 'False', if 'False', it only saves it as a .svg file, default: 'False').
• -ha,--halo: Use halo for ADP ('True' or 'False', default: 'False').
• -z, --zvalue: Z value for ADP (float, default: 3.5).
• -wt, --writetrajs: Write a trajectory file for each cluster ('True' or 'False', default: 'False').
• -wf,--writefreq: Writting frequence (for --writetrajs/-wt option, intgeter, default: 1).
• -nj,--njobs: Number of threads for ADP calculation (integer, default: 1).
• -s ,--slice: Analize a slice of the data (two integers, not used by default, frame count starts with zero).
• -rc, --randomchoice: Make a random choice of --randomchoice/-rc frames (integer, not used by default).
• -h, --help: Show help message and exit.

The dihelist file can be provided in two formats: A single-column file, containing the number of the atoms forming a "chain" of dihedral angles, or a four-columng file, in which the mask of each dihedral is specified. In every case, the atom count starts with 0.

Single-column example:

5
7
8
9
4

Four-column example:

5 7 8 9
7 8 9 4

In both cases, the dihedral angles that will be considered are: 5-7-8-9 and 7-8-9-4.

If the format is 'dihe', a file with NDIHE 1 columns should be provided, containing tha time evolution of NDIHE dihedral angles (columns 2 to NDIHE 1) and a column containing the frame number.

Example:

1 dihe_1_value_in_frame_1 dihe_1_value_in_frame_1 dihe_3_value_in_frame_1 ...
2 dihe_1_value_in_frame_2 dihe_1_value_in_frame_2 dihe_3_value_in_frame_2 ...
3 dihe_1_value_in_frame_3 dihe_1_value_in_frame_3 dihe_3_value_in_frame_3 ...
.
.
.

This option is usefull for saving-time purposes only. Also, a file in this format will be generated as output (with the name 'dihetraj.dat') if format is 'netcdf' of xyz.

The following example performs the analysis for a trajectory file, using only the first 10000 frames, and writes the frames belonging to each cluster as new trajectory files (only 1 out of 10 frames):

python D-clust.py -i trajectory.nc -f netcdf -d dihelist.txt -wt True -wf 10 -s 0 10000

The output files depend on the input files. For example, all the options that involve writing coordinates are only intented for the 'netcdf' format. Otherwise, the will be ignored.

Several messages will be printed on screen during the execution. These include a summary of the clustering processings, as well as info regarding the content of each generated file. Briefly, the files generated will be text files containing the indexes of the frames belonging to each cluster and, if format is 'netcdf', trajectory files with the centers of each (and of each frame, if writetrajs is 'True').

A script for extracting the frames of a netcdf file from a list is also provided (see the tools folder). It can be easily used by typing:

get_frames_from_list.py [-h] -i INPUT -l LIST -o OUTPUT

Where INPUT is a netcdf file, LIST is a list with indexes (that could correspond to the output of the D-clust.py script), and OUTPUT is the name you choose for the netcdf file containing the indexes of INPUT listed in LIST.

Please cite the corresponding DADApy and ADP papers in case you use this tool.