Quickstart¶

Get Started with NetMD¶

To use NetMD with your raw MD data, begin with these steps. A more detailed, example-based quickstart is available on the Quickstart by Example page.

1. Get MD Contacts:¶

First you need to extract residue-residue contacts from MD trajectories with the Python package GetContacts.

(env) $ python3 get_dynamic_contacts.py --topology topology_0.pdb \
                                --trajectory trajectory_0.dcd \
                                --itypes all \
                                --output full_contacts_0.tsv

This will produce a tsv file containing the residue-residue contacts for your trajectory.

For more information refer to the GetContacts GitHub page.

Note

Assign a common prefix to all contact files (e.g. Replica1, Replica2, …), as it will be used in the output generated by NetMD.

2. Launch NetMD:¶

Prepare to launch NetMD. To give the input files you can either list the files in the command line or specify a directory tree and a prefix for recursive exploration:

Sequence of Files

(env) $ netmd -F [FullReplica1_WT.tsv FullReplica4_WT.tsv ...] -o ./results --verbose

NetMD will iterate over the list of files and generate the embeddings for each one.

Directory Tree

(env) $ netmd -I example_dir FullReplica --verbose

NetMD will recursivly explore the directory tree starting from example_dir and generate the embeddings for each file with the prefix FullReplica.

Note

The -I option will only work if the input files are in the same directory tree. If you have files in different directories, you can list them with the -F option.

That's it, you are ready to launch NetMD! check

However, the program will use default values for many of its options. If you have the time, please take a look at NetMD’s help section to learn more about the available options.

3. Get a Coffee:¶

Depending on the number of frames in your trajectory, the embedding process may take a few minutes. So, while you wait, why not grab a cup of coffee?

4. Analyze Results:¶

The final output will include:

Embeddings for each MD replica frame.
The filtered graph representation used to generate the embeddings.
The dynamic time warping (DTW) mapping between the frames and the barycenter.
The iterative pruning based on the replicas’ distance from the barycenter.
Several plots to help visualize and understand results and guide further analysis.

To learn more about the output files and how to interpret them, refer to the Quickstart by Example section.