This project enhances the SNR of Cryo-EM images, using 2-D classification and expectation-maximization.
python setup.py sdist
pip install dist/cryo_signal_enhance-1.0.tar.gz
The package recieves a mandatory starfile (.star) as its input.
The starfile should be pointing to images of particle (post particle-picking)
under the field _rlnImageName.
For CTF correction, the starfile should also include the following parameters:
- _rlnAmplitudeContrast
- _rlnSphericalAberration
- _rlnDefocusAngle
- _rlnDefocusV
- _rlnDefocusU
- _rlnVoltage
The starfile is given using the following
--starfile starfile_path
The output is an mrcs (.mrcs) file containing the images with enhanced SNR, and is given by
--output output_path
Additional optional arguments are:
--basis_file_out Write sPCA basis to file, allowing to skip preprocessing next time (.pkl)
--basis_file_in Optional input file for sPCA basis. When given, skips preprocessing (.pkl)
--N Number of images to take from starfile
--verbose Print progress to screen
--downsample Preprocessed image size
--batch_size Batch size for PSD noise estimation
--num_coeffs Number of sPCA coefficient
--class_theta_step Sampling rate for the in-plane angle
--num_classes Number of classes to find (<= N)
--class_size Number of images in each class (<= N)
--class_gpu Use available gpu for matrix calculations (recommended)
-1 = don't use gpu
0 = select gpu with most available memory
1,2,... = use specific gpu
--ctf Perform CTF correction
--iter Maximum number of iterations to perform
--norm Perform normalisation-error correction
--em_num_inputs Number of images to use for EM (taken from best of each class) (<= class_size)
--em_par Perform EM parallelly (recommended)
--num_class_avg Number of class averages to produce (<= num_classes)
cryo_signal_enhance --starfile dir/some_star.star --output dir/class_averages.mrcs
cryo_signal_enhance --starfile dir/some_star.star --output dir/class_averages.mrcs --num_classes 3000 --num_class_avg 2000