a python script to remove particles for re-weighting angular distribution.
keywords: Cryo-Em,relion,relin3.1,angular distribution,Euler angle
Please edit variables below in the main Function before executing the script.
- Change 'star_file' as the original star file.(only relion3.1 run_data.star from Refine3D tested)
- Change 'output_dir' for multiple star file generation. If output_dir does not exist, a new dir will be created.
- Jupyter lab will be helpful.
input:
star_file:path-like str
output:
final_data_list:
list[(int interval indices,
pandas.Interval rot_interval,
pandas.Interval tilt_interval,
int number of particles in this group,
list[int] particle indices in this group,)]
output_dir: path-like str
'parse_and_preview' will generate several plots for angular Distribution. * relion_Rot vs number of particles * relion_Tilt vs number of particles * relion_Psi vs number of particles * heatmap of relion_Rot - relion_Tilt - number of particles plots will be saved as png in the same dir as 'star_file'
a json file will also be generated to save list[(Rot,Tilt,Psi,particle-name:str)] parsed from run_data.star.
input:
star_file:path-like str. original run_data.star.
output_dir:path-like str.
final_data_list=final_data_list.
rot_min:float range=-200.0.
rot_max:float range=+200.0.
tilt_min:float range=-1.0.
tilt_max:float range=+ 200.0.
remove_percent:float range=0.3.
select_percent:float range=0.2. # when auto_mode=False,this value will be ignored
particle_number_threshold=-1.
# bigger than this value is selected for next removal step
auto_mode=False,
# 'normal-mode' and 'auto-mode' is two strategies to remove particles for re-weighting angular distribution.
output:
new star file for Relion3.1
plot: sorted 'Euler angle intervals vs number of particles' before removing paritcles.
plot: sorted 'Euler angle intervals vs number of particles' after removing paritcles.
brief report of total particles number and removed particle number.
After 'parse_and_preview' you can check angular Distribution from the heatmap of relion_Rot - relion_Tilt - number of particles. Then, 'normal-mode' will remove particles in range(rot_min,rot_max) and range(tilt_min,tilt_max) by a remove_percent. And particle_number_threshold will filter out sections with particle numbers bigger than that threshold for removing.
For example, when Rot_min=10,Rot_max=60,Tilt_min=10,Tilt_max=60,remove_percent=0.3,particle_number_threshold=1700, script will delete 30% particles randomly for each section, of which particle number is larger than 1700 and section in range_rot(10,60),range_tilt(10,60).
Another round of 'parse_and_preview' with the new star file will be helpful to compare the angular distribution before and after removal.
When using 'auto-mode', you should input True Flag in Function 'fix_euler_distribution'. Instead of inputting an Euler angle range to select sections, all sections (all intervals in the heatmap) will be sorted by the number of particles. 'select_percent' will select sections from the sections with most particles to the least until the selected particle number reach the ('select_percent' x total particle number). Then, for each selected section, random particles will be deleted by 'remove_percent'.
By using the output star file as input, you are able to check angular Distribution conviniently from the heatmap to tell whether to remove more or not.
