This repository contains a script for utilizing averaging methods in MODELLER for protein structure prediction. The script is designed to enhance the accuracy and reliability of predicted protein structures through model averaging. Authors
Command-line Arguments
The script can be executed from the command line with the following options:
--alnfile: Path to the alignment file (e.g., 'tar_tem_alignment.ali').
--knowns: Target structure name in the alignment file (e.g., '1bdm').
--sequence: Sequence name in the alignment file (e.g., 'TvLDH').
Optional Arguments
--n_jobs (default: 4): Number of parallel jobs to use during model averaging.
--n_runs (default: 24): Number of MD runs to perform.
--model_name (default: '_averaged_model'): Name of the output model file.
--scoring_func (default: 'dope'): Scoring function for model evaluation.
--select_top (default: 0.1): Fraction of top decoys to be used for averaging.
Example Usage
-
Standard usage
python altmod_run_averaging.py --alnfile tar_tem_alignment.ali --knowns 1bdm --sequence TvLDH -
Rescoring option
python altmod_run_averaging.py --alnfile 3bwm_4pyi_5zw4.ali --knowns 4pyi,5zw4 --sequence 3bwm --lig_rescoring True
Notes
-
The script initializes an 'automodel' object using the Automodel_averaging class.
-
Additional parameters, such as md_level, can be adjusted directly within the script if needed.
-
The final averaged model will be written to the specified output file.
-
If you want to build multiple averaged models for the same protein, remember to change the random seed of MODELLER within the script, by adding the lines:
random_seed = 42 env.io.random_seed = random_seed