[Standard Relaxation Procedure] Clarification Needed

[Zehui127]

[Standard Relaxation Procedure] Clarification Needed

[lorenzo-rovigatti]

Hi! I can't remember on top of my head what oxdna.org does, but maybe @zoombya or @ErikPoppleton do.
About your general question, relaxation depends on your initial structure, and my experience is that it is hard to find a reliable protocol that

1. is as efficient as possible
2. works every time

Most of the times you'll need some sort of supervision (or to check some heuristics that let your code evaluate whether more relaxation is required).

[ErikPoppleton]

The relaxation protocol on oxdna.org uses the files from oxDNA/analysis/example_input_files (actually I think it runs the simulations a bit longer than those files for consistency). We have found that this procedure works for almost all structures. Depending on how good your initial configuration is, you might need to adjust the length of MD relax step or modify the max_backbone_force parameter to get good relaxation. One thing we strongly recommend is to generate an external force file with relatively strong forces (we arbitrarily use 3.1) holding all pre-existing base pairs together as relaxing stretched bonds can tear your initial structure apart.

The procedure was first described in section 3.2 of this protocol. We discuss it further in section 4 of this protocol.

[zoombya]

Erik's answer is extremely comprehensive, additionally I wanted to add that it makes little sense to use
save_dat(config, OUTPUT_DIR) \ since you already save while oxDNA is running.
OAT has the minify script to handle trajectories, and if you want to convert them to PDB afterwards you would need the raw data anyway.
Additionally the boilerplate Simulation class provides methods to visualize even trajectories, doing the minify step you copied from the library.