MD DataRecord a brief overview¶
Molecular Dynamics (MD) simulations are notoriously time consuming and computational demanding. In terms of data, a lot of information need to be stored and retrieved such as atomic coordinates, atomic velocities, forces and energies just to cite only few; extracting and managing this data then becomes crucial. The MD DataRecord API simplifies the access to the md data in the MD floe programming for Orion. In the OpenEye Datarecord model, data is exchanged between cubes in format of data records where POD data, custom objects, json data etc. are stored and retrieved by using the associated field names and types. The MD Datarecord API is built on the top of the OpenEye Datarecord, standardizing the record content data produced during MD runs in a well-structured format and providing an API point to its access.
MD DataRecord structure¶
The MD data produced along MD runs is structured in what is named the md record:
the md record contains a sub-record named md stages where md information is saved. This sub-record is a list of md stage records;
each md stage is a record itself with an associated name, type, log data, topology and MD State info. The latter is a custom object that stores data useful to restart MD runs such as, atomic positions, velocities and box information;
the md record at the top level contains a Parmed object used to carry the whole system parametrization data
the md record at the top level can also contain other data such as the MDComponents object used to carry info related to the different md system parts such as ligand, protein, solvent, cofactors etc. or can contains the starting ligand and protein with their names, unique identifiers such as the flask id , cofactor ids etc.
The following picture shows the md record structure and its main components
The md record is user accessible by using the MDDataRecord API. In order to use it, the MDOrion package must be installed. The installation of the package also requires to have access to the OpenEye Magpie repository for some dependencies. The API has been designed to transparently work locally and in Orion. To use the API the user needs to create a MDDatarecord object starting from an OERecord object, after that getter and setter functions can be used to access the md data (see full MDDatatecord API Documentation).
The following code snippets give an idea on how to use the API.
In the following examples record is an OpenEye Datarecord produced running MD floes such as Solvate and Run MD or Solvate and Run Protein-Ligand MD. Starting from the MDOrion pkg v2.0.0, the datasets produced from the Short Trajectory MD with analysis floe is “ligand centric” and the MDDatarecord API cannot be directly applied to the produced records. However, the API still works on the conformer (poses) ligand sub-records which are still md records
from orionmdcore.mdrecord import MDDataRecord # To use the MD Datarecord API an MDDataRecord instance is built starting from an OERecord md_record = MDDataRecord(record) # At this point getters and setters can be used to # extract/set info from/to the md record # MD Stage names available stage_names = md_record.get_stages_names # Get a MD Record Stage md_stage_production = md_record.get_stage_by_name(stage_names) # Extract the logging info from a stage info_stage = md_record.get_stage_info(stg_name=stage_names) # Extract the MD State from a stage md_set_up_state = md_record.get_stage_state(stg_name=stage_names) # Extract the Parmed Structure from the md record and synchronize the # positions, velocities and box data to the selected stage state pmd_structure = md_record.get_parmed(sync_stage_name=stage_names) # Extract the OEMol system flask from the record and its title flask = md_record.get_flask flask_title = md_record.get_title # Extract the trajectory file name associated with a stage. In this # case the stage trajectory in unpacked and the trajectory name # can be used in any md analysis pkg to be loaded trj_name = md_record.get_stage_trajectory(stg_name=stage_names) # Add a new Stage to the md_stages record md_record.add_new_stage(stage_name="New_Stage", stage_type="NPT", topology=flask, mdstate=md_set_up_state, data_fn="test.tar.gz")
MDDatatecord API Documentation¶
Follow the API documentation