CryoDRGN

https://ez-lab.gitbook.io/cryodrgn
Admin
CryoDRGN is a neural network-based software package for heterogeneous 3D reconstruction of protein...
Tokens:35,200
Snippets:298
Trust Score:9.8
Update:6 months ago
Show doc for...
Tokens:
Raw
### Install Dependencies and Source Code

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Manual installation of PyTorch and cloning the repository for source-based installation.

```bash
(cryodrgn) $ conda install pytorch=2.9 cudatoolkit=12.9 -c pytorch
```

```bash
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 590.48.01              Driver Version: 590.48.01      CUDA Version: 13.1     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA A100-SXM4-80GB          On  |   00000000:01:00.0 Off |                    0 |
| N/A   36C    P0             73W /  500W |    8004MiB /  81920MiB |      0%      Default |
...
```

```bash
# Clone source code and install
(cryodrgn) $ git clone https://github.com/ml-struct-bio/cryodrgn.git
(cryodrgn) $ cd cryodrgn
(cryodrgn) $ pip install .
```

```bash
(cryodrgn) $ unzip cryodrgn-3.5.4.zip
(cryodrgn) $ cd cryodrgn-3.5.4
(cryodrgn) $ pip install .
```

--------------------------------

### Install Beta cryoDRGN

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Installation of the early beta version from the TestPyPI channel.

```bash
(cryodrgn) $ pip install -i https://test.pypi.org/simple/ \
                         --extra-index-url https://pypi.org/simple/ \
                         cryodrgn --pre
```

--------------------------------

### Install cryoDRGN via Pip

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Standard pip installation command.

```bash
(cryodrgn) $ pip install cryodrgn
```

--------------------------------

### Install core dependencies

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Installs PyTorch and pandas into the active environment.

```bash
conda install pytorch -c pytorch
conda install pandas
```

--------------------------------

### Run CryoDRGN Installation Test Script

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Navigate to the source code directory and execute the quicktest.sh script to verify all dependencies are installed correctly. This script runs a small dataset and should complete within approximately 20 seconds.

```bash
(cryodrgn) $ cd [sourcecode directory]/tests
(cryodrgn) $ ./quicktest.sh
```

--------------------------------

### Clone and install CryoDRGN

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Downloads the source code from GitHub and installs the package.

```bash
git clone https://github.com/ml-struct-bio/cryodrgn.git
cd cryodrgn
git checkout 4.2.0 # or desired version
pip install .
```

--------------------------------

### Start Jupyter Notebook Locally

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Use this command to start a Jupyter notebook server on the local machine where CryoDRGN is installed. It will provide a URL to access the notebook in your browser.

```bash
(cryodrgn) $ jupyter notebook
[I 14:47:21.672 NotebookApp] JupyterLab extension loaded from /Users/zhonge/anaconda3/envs/cryodrgn/lib/python3.7/site-packages/jupyterlab
[I 14:47:21.672 NotebookApp] JupyterLab application directory is /Users/zhonge/anaconda3/envs/cryodrgn/share/jupyter/lab
[I 14:47:21.674 NotebookApp] Serving notebooks from local directory: /Users/zhonge/dev/cryodrgn/master
[I 14:47:21.674 NotebookApp] Jupyter Notebook 6.1.4 is running at:
[I 14:47:21.674 NotebookApp] http://localhost:8888/?token=6e54c833c7da8c8c6e7c418feeb0dbbc76b0752a82d114f3
[I 14:47:21.674 NotebookApp]  or http://127.0.0.1:8888/?token=6e54c833c7da8c8c6e7c418feeb0dbbc76b0752a82d114f3
[I 14:47:21.674 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 14:47:21.689 NotebookApp] 

    To access the notebook, open this file in a browser:
        file:///Users/zhonge/Library/Jupyter/runtime/nbserver-67879-open.html
    Or copy and paste one of these URLs:
        http://localhost:8888/?token=6e54c833c7da8c8c6e7c418feeb0dbbc76b0752a82d114f3
     or http://127.0.0.1:8888/?token=6e54c833c7da8c8c6e7c418feeb0dbbc76b0752a82d114f3

```

--------------------------------

### CryoDRGN Training Command Example

Source: https://ez-lab.gitbook.io/cryodrgn/installation

This is an example of the command executed by the quicktest.sh script for training a VAE model. It specifies input data, output directory, learning rate, seed, pose information, and latent space dimensions.

```bash
cryodrgn train_vae data/hand.mrcs -o output/toy_recon_vae --lr .0001 --seed 0 --poses data/hand_rot.pkl --zdim 10 --pe-type gaussian
```

--------------------------------

### Quick Install cryoDRGN

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Standard installation procedure using pip within a dedicated conda environment.

```bash
# We recommend first creating a clean anaconda environment
(base) $ conda create --name cryodrgn python=3.13
(base) $ conda activate cryodrgn

# Install cryodrgn
(cryodrgn) $ pip install cryodrgn

# Check version
(cryodrgn) $ cryodrgn --version
```

--------------------------------

### Install cryoDRGN environment

Source: https://ez-lab.gitbook.io/cryodrgn

Create a dedicated conda environment and install the cryoDRGN package.

```bash
$ conda create --name cryodrgn-env python=3.13
$ conda activate cryodrgn-env
(cryodrgn-env) $ pip install cryodrgn
```

--------------------------------

### Run cryoDRGN-AI ab initio reconstruction

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-ai-ab-initio-reconstruction

Example command for running ab initio volume reconstruction. Requires particle stack, CTF parameters, an output directory, and the latent space dimension.

```bash
cryodrgn abinit particles_128.mrcs --ctf ctf.pkl -o 001_drgnai_abinit_zdim.4/ --zdim 4
```

--------------------------------

### Example cryoDRGN analysis log

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Sample output log showing the execution of the analysis command and subsequent volume generation.

```text
(cryodrgn4) [Thu Feb 04 02:57 00] cryodrgn analyze . 50
2021-02-04 02:57:18     Saving results to /nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/analyze.50
2021-02-04 02:57:18     Perfoming principal component analysis...
2021-02-04 02:57:18     Explained variance ratio:
2021-02-04 02:57:18     [0.22400902 0.2087886  0.16459696 0.10960732 0.09572161 0.07636773
 0.06840914 0.05249962]
2021-02-04 02:57:18     Generating volumes...
2021-02-04 02:57:18     Running command:
cryodrgn eval_vol /nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/weights.50.pkl --config /nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/config.yaml --zfile /nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/analyze.49/pc1/z_values.txt -o /nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/analyze.49/pc1 --Apix 1
2021-02-04 02:57:20     Use cuda True
2021-02-04 02:57:20     Namespace(Apix=1.0, D=None, activation='relu', config='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/config.yaml', domain=None, downsample=None, enc_mask=None, encode_mode=None, flip=False, func=<function main at 0x7ffed8e3fe18>, l_extent=None, n=10, norm=None, o='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/analyze.49/pc1', pdim=None, pe_dim=None, pe_type=None, players=None, prefix='vol_', qdim=None, qlayers=None, verbose=False, weights='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/weights.49.pkl', z=None, z_end=None, z_start=None, zdim=None, zfile='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/tutorial/00/analyze.49/pc1/z_values.txt')
2021-02-04 02:57:20     Loaded configuration:
{'dataset_args': {'ctf': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/ctf.new.pkl',
                    'datadir': None,
                    'do_pose_sgd': False,
                    'ind': None,
                    'invert_data': True,
                    'keepreal': False,
                    'norm': [0, 278.59372],
                    'particles': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/128/projections.128.mrcs',
                    'poses': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/pose.pkl',
                    'window': True},
   'lattice_args': {'D': 129, 'extent': 0.5, 'ignore_DC': True},
   'model_args': {'activation': 'relu',
                  'domain': 'fourier',
                  'enc_mask': 64,
                  'encode_mode': 'resid',
                  'pdim': 256,
                  'pe_dim': None,
                  'pe_type': 'geom_lowf',
                  'players': 3,
                  'qdim': 256,
                  'qlayers': 3,
                  'zdim': 8},
   'seed': 43266}
2021-02-04 02:57:29     Using circular lattice with radius 64
2021-02-04 02:57:30     Generating 10 volumes
2021-02-04 02:57:30     [-0.46911267 -0.69718513 -0.0095188  -2.78874987  0.1540933   0.16372231
    0.31355665  2.17943362]
2021-02-04 02:57:32     [-0.39820911 -0.56545639 -0.01828557 -2.24070284  0.12997294  0.12582034
    0.25060696  1.75664491]
2021-02-04 02:57:33     [-0.32730555 -0.43372764 -0.02705235 -1.69265581  0.10585258  0.08791838
    0.18765726  1.33385621]
2021-02-04 02:57:34     [-0.25640199 -0.3019989  -0.03581912 -1.14460877  0.08173222  0.05001641
    0.12470757  0.91106751]
2021-02-04 02:57:35     [-0.18549843 -0.17027016 -0.0445859  -0.59656174  0.05761186  0.01211444
    0.06175788  0.48827881]
2021-02-04 02:57:36     [-0.11459487 -0.03854141 -0.05335267 -0.04851471  0.0334915  -0.02578752
   -0.00119181  0.06549011]
2021-02-04 02:57:36     [-0.04369131  0.09318733 -0.06211945  0.49953232  0.00937114 -0.06368949
   -0.0641415  -0.35729859]
```

--------------------------------

### Metadata File Content

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Example content of the Parameters.star file showing particle image names and CTF parameters.

```text
data_

loop_
_rlnImageName #1
_rlnMicrographName #2
_rlnDefocusU #3
_rlnDefocusV #4
_rlnDefocusAngle #5
_rlnVoltage #6
_rlnSphericalAberration #7
_rlnAmplitudeContrast #8
_rlnMagnification #9
_rlnDetectorPixelSize #10
1@L17Combine_weight_local.mrcs 1 15301.1 14916.4 5.28  300 2.7 0.07 38168 5
2@L17Combine_weight_local.mrcs 2 15303.0 14918.4 5.28  300 2.7 0.07 38168 5
3@L17Combine_weight_local.mrcs 3 15150.7 14766.0 5.28  300 2.7 0.07 38168 5
4@L17Combine_weight_local.mrcs 4 15181.9 14797.3 5.28  300 2.7 0.07 38168 5
5@L17Combine_weight_local.mrcs 5 15366.8 14982.2 5.28  300 2.7 0.07 38168 5
6@L17Combine_weight_local.mrcs 6 15305.0 14920.3 5.28  300 2.7 0.07 38168 5
7@L17Combine_weight_local.mrcs 7 15177.4 14792.7 5.28  300 2.7 0.07 38168 5
```

--------------------------------

### Install visualization dependencies

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Installs libraries and Jupyter extensions required for latent space visualization.

```bash
conda install seaborn scikit-learn 
conda install umap-learn jupyterlab ipywidgets cufflinks-py "nodejs>=15.12.0" -c conda-forge
jupyter labextension install @jupyter-widgets/jupyterlab-manager --no-build
jupyter labextension install jupyterlab-plotly --no-build
jupyter labextension install plotlywidget --no-build
jupyter lab build
```

--------------------------------

### Example cryoDRGN analysis log

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

A sample output log showing the principal component analysis process and subsequent volume generation commands.

```bash
2021-02-28 14:24:49     Saving results to /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49
2021-02-28 14:24:49     Perfoming principal component analysis...
2021-02-28 14:24:49     Explained variance ratio:
2021-02-28 14:24:49     [0.21076616 0.1977381  0.11697037 0.09955638 0.0985605  0.09767462
 0.09139399 0.08733987]
2021-02-28 14:24:49     Generating volumes...
2021-02-28 14:24:49     Running command:
cryodrgn eval_vol /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/weights.49.pkl --config /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/config.yaml --zfile /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc1/z_values.txt -o /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc1 --Apix 1
2021-02-28 14:24:50     Use cuda True
2021-02-28 14:24:50     Namespace(Apix=1.0, D=None, activation='relu', config='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/config.yaml', domain=None, downsample=None, enc_mask=None, encode_mode=None, flip=False, func=<function main at 0x2000ac580e18>, l_extent=None, n=10, norm=None, o='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc1', pdim=None, pe_dim=None, pe_type=None, players=None, prefix='vol_', qdim=None, qlayers=None, verbose=False, weights='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/weights.49.pkl', z=None, z_end=None, z_start=None, zdim=None, zfile='/nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc1/z_values.txt')
2021-02-28 14:24:50     Loaded configuration:
{'dataset_args': {'ctf': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/ctf.new.pkl',
                    'datadir': None,
                    'do_pose_sgd': False,
                    'ind': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/aux/ind2.train.pkl',
                    'invert_data': False,
                    'keepreal': False,
                    'norm': [0, 325.65057],
                    'particles': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/256/projections.txt',
                    'poses': '/nobackup/users/zhonge/cryodrgn/11_l17_ribo/data/pose.pkl',
                    'window': False},
   'lattice_args': {'D': 257, 'extent': 0.5, 'ignore_DC': True},
   'model_args': {'activation': 'relu',
                  'domain': 'fourier',
                  'enc_mask': 128,
                  'encode_mode': 'resid',
                  'pdim': 1024,
                  'pe_dim': None,
                  'pe_type': 'geom_lowf',
                  'players': 3,
                  'qdim': 1024,
                  'qlayers': 3,
                  'zdim': 8},
   'seed': 21152}
2021-02-28 14:24:56     Using circular lattice with radius 128
2021-02-28 14:24:56     Generating 10 volumes
2021-02-28 14:24:56     [-0.13908401 -0.66560406  0.95547887 -0.6795193   2.33642701  2.05639807
   -1.21175412 -2.10233111]
2021-02-28 14:25:06     [-0.13527223 -0.48584066  0.75453086 -0.51289588  1.81356009  1.58481893
   -0.9511668  -1.6463648 ]
2021-02-28 14:25:15     [-0.13146044 -0.30607727  0.55358285 -0.34627247  1.29069317  1.11323979
   -0.69057947 -1.19039849]
2021-02-28 14:25:23     [-0.12764866 -0.12631387  0.35263484 -0.17964905  0.76782624  0.64166065
   -0.42999215 -0.73443218]
2021-02-28 14:25:32     [-0.12383687  0.05344952  0.15168683 -0.01302564  0.24495932  0.1700815
   -0.16940482 -0.27846588]
2021-02-28 14:25:41     [-0.12002508  0.23321291 -0.04926117  0.15359777 -0.2779076  -0.30149764
    0.09118251  0.17750043]
2021-02-28 14:25:49     [-0.1162133   0.41297631 -0.25020918  0.32022119 -0.80077453 -0.77307678
    0.35176983  0.63346674]
2021-02-28 14:25:58     [-0.11240151  0.5927397  -0.45115719  0.4868446  -1.32364145 -1.24465592
    0.61235716  1.08943305]
2021-02-28 14:26:07     [-0.10858973  0.7725031  -0.6521052   0.65346802 -1.84650837 -1.71623506
    0.87294448  1.54539936]
2021-02-28 14:26:15     [-0.10477794  0.95226649 -0.85305321  0.82009143 -2.3693753  -2.1878142
    1.13353181  2.00136567]
2021-02-28 14:26:24     Finsihed in 0:01:33.787293
2021-02-28 14:26:25     Running command:
cryodrgn eval_vol /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/weights.49.pkl --config /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/config.yaml --zfile /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc2/z_values.txt -o /nobackup/users/zhonge/cryodrgn/11_l17_ribo/04_vae256_gpu/3_gpu1_b8/analyze.49/pc2 --Apix 1
2021-02-28 14:26:27     Use cuda True
```

--------------------------------

### Monitor cryoDRGN training logs

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Example of the output format generated during the training process, showing epoch progress and loss metrics.

```text
2021-02-03 17:48:14     # [Train Epoch: 1/50] [2000/131899 images] gen loss=1.039896, kld=7.318097, beta=0.125000, loss=1.039967
2021-02-03 17:48:17     # [Train Epoch: 1/50] [3000/131899 images] gen loss=1.010868, kld=8.658796, beta=0.125000, loss=1.010953

.
.
.

2021-02-04 00:39:24     # [Train Epoch: 50/50] [126000/131899 images] gen loss=0.989016, kld=26.381516, beta=0.125000, loss=0.989272
2021-02-04 00:39:28     # [Train Epoch: 50/50] [127000/131899 images] gen loss=0.948055, kld=24.869091, beta=0.125000, loss=0.948297
2021-02-04 00:39:31     # [Train Epoch: 50/50] [128000/131899 images] gen loss=1.017258, kld=25.462738, beta=0.125000, loss=1.017505
2021-02-04 00:39:35     # [Train Epoch: 50/50] [129000/131899 images] gen loss=1.033358, kld=30.238625, beta=0.125000, loss=1.033653
2021-02-04 00:39:38     # [Train Epoch: 50/50] [130000/131899 images] gen loss=0.966860, kld=24.983553, beta=0.125000, loss=0.967103
2021-02-04 00:39:41     # [Train Epoch: 50/50] [131000/131899 images] gen loss=0.982898, kld=25.034485, beta=0.125000, loss=0.983141
2021-02-04 00:39:44     # =====> Epoch: 50 Average gen loss = 0.998177, KLD = 26.589939, total loss = 0.998435; Finished in 0:07:27.660706
2021-02-04 00:41:18     Finsihed in 6:55:17.244481 (0:08:18.344890 per epoch)
```

--------------------------------

### CryoDRGN Graph Traversal with Assembly State Anchors

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Example of using `cryodrgn graph_traversal` with specific particle indices representing assembly states as anchor points to generate a trajectory.

```bash
$ cryodrgn graph_traversal ../z.49.pkl --anchors 67124 19716 21262 13647 34996 14564 57647 -o assembly_path_A/path.txt --out-z assembly_path_A/z.path.txt

```

--------------------------------

### CryoDRGN Graph Traversal Example

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Example usage of `cryodrgn graph_traversal` to find a path between anchor points (kmeans cluster centers) in a latent space embedding file.

```bash
$ cd analyze.49 # assuming you've already run cryodrgn analyze
$ cryodrgn graph_traversal ../z.49.pkl --anchors $(cat kmeans20/centers_ind.txt) -o graph_traversal/path.txt --out-z graph_traversal/z.path.txt

```

--------------------------------

### Example: Extract Poses from CryoSPARC File

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

This command demonstrates how to extract poses from a cryoSPARC .cs file using cryodrgn parse_pose_csparc, specifying the input file, reconstruction box size (-D), and output .pkl file (-o).

```bash
$ cryodrgn parse_pose_csparc cryosparc_P4_J33_004_particles.cs -D 320 -o poses.pkl
```

--------------------------------

### Create ChimeraX Movie from Volumes

Source: https://ez-lab.gitbook.io/cryodrgn/making-long-trajectories-with-cryodrgn-graph_traversal

Commands to visualize the generated density maps in ChimeraX and record a movie. This involves loading volumes, setting colors, starting recording, and encoding the movie.

```text
vol all level XYZ
vol all color cornflowerblue
mov record
mseries all
mov encode graph_traversal.isoXYZ.mp4 framerate 15
```

--------------------------------

### Show cryoDRGN-AI ab initio help

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-ai-ab-initio-reconstruction

Displays the help message for the `cryodrgn abinit` command, outlining available options and usage.

```bash
cryodrgn abinit -h
```

--------------------------------

### View analyze_landscape help

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-conformational-landscape-analysis

Displays all available settings and default values for the analyze_landscape script.

```bash
(cryodrgn) $ cryodrgn analyze_landscape -h
```

--------------------------------

### Display help for write_star utility

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

View the command-line arguments and usage instructions for the write_star tool.

```bash
(cryodrgn) $ cryodrgn_utils write_star -h
usage: cryodrgn_utils write_star [-h] [--ctf CTF] [--poses POSES] [--ind IND] [--full-path] -o O particles

Create a Relion 3.0 star file from a particle stack and ctf parameters

positional arguments:
  particles            Input particles (.mrcs, .txt, .star)

optional arguments:
  -h, --help           show this help message and exit
  --ctf CTF            Optionally include ctf information from a .pkl file
  --poses POSES        Optionally include pose information from a .pkl file
  --ind IND            Optionally filter by selected index array (.pkl)
  --full-path          Write the full path to particles (default: relative paths)
  -o O                 Output .star file
```

--------------------------------

### Install specific cryoDRGN version

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-user-guide/news-and-release-notes

Use this command to install a specific version of the cryoDRGN package via pip.

```bash
pip install 'cryodrgn==3.5.4'
```

--------------------------------

### CryoDRGN Training Workflow Step 1

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Initial training pass on lower resolution images (e.g., D=128) with the default architecture. This step is useful for sanity checking results and removing junk particles.

```bash
First, train on lower resolution images (e.g. D=128) using the default architecture (fast) as an initial pass to sanity check results and remove junk particles. **(D=128, smaller 256x3 architecture)**
```

--------------------------------

### Display cryoDRGN help commands

Source: https://ez-lab.gitbook.io/cryodrgn

Access help documentation for the main cryoDRGN and utility executables.

```bash
(cryodrgn-env) $ cryodrgn -h
(cryodrgn-env) $ cryodrgn_utils -h
```

--------------------------------

### Run initial ab initio training

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-ai-ab-initio-reconstruction/cryodrgn-ai-ab-initio-empiar-10076-tutorial

Executes the cryodrgn abinit command to perform an initial pass on downsampled particles.

```bash
cryodrgn abinit particles.128.mrcs --zdim 4 --lazy --ctf ctf.pkl -o 50S_abinit/001_defaults.128/ --uninvert-data
```

--------------------------------

### Check CryoDRGN Version

Source: https://ez-lab.gitbook.io/cryodrgn/helpful-resources

Use this command to check the installed version of CryoDRGN. Ensure you have the correct version for compatibility.

```bash
cryodrgn --version
```

--------------------------------

### Access cryodrgn analyze help

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Displays the command-line interface help for the cryodrgn analyze tool.

```bash
(cryodrgn) $ cryodrgn analyze -h
```

--------------------------------

### Display CryoDRGN Help Information

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Run this command to see the available cryodrgn commands and options. Ensure the cryodrgn executable is in your PATH or activated environment.

```bash
(cryodrgn) $ cryodrgn -h
```

--------------------------------

### Update CryoDRGN Installation

Source: https://ez-lab.gitbook.io/cryodrgn/installation

To update CryoDRGN, checkout the desired version and reinstall using pip. Create a new environment for parallel versions.

```bash
(cryodrgn) $ cd /path/to/git/repo/with/source/code
(cryodrgn) $ git checkout 3.5.0  # or `git pull origin main` to get the latest
(cryodrgn) $ pip install .
```

--------------------------------

### Inspect Metadata File

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Display the first 20 lines of the Parameters.star metadata file to verify CTF parameters.

```bash
$ head Parameters.star -n 20
```

--------------------------------

### Download Dataset via Aspera

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Use the ascp command-line tool to download the EMPIAR-10076 dataset.

```bash
$ ascp -QT -l 200M -P33001 -i ~/.aspera/connect/etc/asperaweb_id_dsa.openssh emp_ext3@hx-fasp-1.ebi.ac.uk:/10076 .
```

--------------------------------

### Train CryoDRGN-ET VAE Model

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-et-subtomogram-analysis

Train the Variational Autoencoder (VAE) model for cryoDRGN-ET using prepared input files. The `--encode-mode tilt` is essential for tilt series data. A KL regularization weight (`--beta`) of 0.025 is recommended as a starting point.

```bash
(cryodrgn) $ cryodrgn train_vae particles_2d.128.star --ctf ctf.pkl --poses pose.pkl  --encode-mode tilt --dose-per-tilt 3.0 --zdim 8 -n 50 --beta 0.025 -o 01_trainvae_128/
```

--------------------------------

### CryoDRGN Training Workflow Step 3

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Final training on the full resolution image stack (up to D=256) with a large architecture, following validation, pose optimization, and particle filtering.

```bash
Finally, after validation, pose optimization, and any necessary particle filtering, then train on the full resolution image stack (up to D=256) with a large architecture. **(D=256, larger 1024x3 architecture)**
```

--------------------------------

### Create and Activate Conda Environment

Source: https://ez-lab.gitbook.io/cryodrgn/installation

Commands to initialize the environment for cryoDRGN.

```bash
(base) $ conda create --name cryodrgn python=3.13
```

```bash
(base) $ conda activate cryodrgn
(cryodrgn) $
```

--------------------------------

### Set up SSH Tunnel for Remote Jupyter Notebook Access

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

These commands are used to access a Jupyter notebook running on a remote machine from your local machine. First, start the notebook server on the remote machine without a browser, then establish an SSH tunnel from your local machine to the remote server.

```bash
# on the remote machine, start the jupyter notebook server in the cryodrgn workdir
(cryodrgn) $ jupyter notebook --no-browser --port 8888 # specify an arbitrary port number > 1024

# on your local machine, open a unix terminal (or MS-DOS on Windows) and type the following command to set up the SSH tunnel
$ ssh -N -f -L localhost:8888:localhost:8888 remote_username@remote_host_name # replace remote_username and remote_host_name with your login information

# then navigate to a browser and type in localhost:8888 in the address bar
```

--------------------------------

### Configure GPU training for cryoDRGN

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Commands to execute training with single or multiple GPUs using the --multigpu flag and CUDA_VISIBLE_DEVICES environment variable.

```bash
$ cryodrgn train_vae ... # Run on GPU 0
$ cryodrgn train_vae ... --multigpu # Run on all GPUs on the machine
$ CUDA_VISIBLE_DEVICES=0,3 cryodrgn train_vae ... --multigpu # Run on GPU 0,3
```

--------------------------------

### CryoDRGN Graph Traversal Help

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Displays the help message for the `cryodrgn graph_traversal` command, outlining its arguments and options.

```bash
(cryodrgn) $ cryodrgn graph_traversal -h
usage: cryodrgn graph_traversal [-h] --anchors ANCHORS [ANCHORS ...]
                                  [--max-neighbors MAX_NEIGHBORS]
                                  [--avg-neighbors AVG_NEIGHBORS]
                                  [--batch-size BATCH_SIZE]
                                  [--max-images MAX_IMAGES]
                                  -o PATH.TXT --out-z
                                  Z.PATH.TXT
                                  data

Find shortest path along nearest neighbor graph

positional arguments:
  data                  Input z.pkl embeddings

optional arguments:
  -h, --help            show this help message and exit
  --anchors ANCHORS [ANCHORS ...]
                        Index of anchor points
  --max-neighbors MAX_NEIGHBORS
  --avg-neighbors AVG_NEIGHBORS
  --batch-size BATCH_SIZE
  --max-images MAX_IMAGES
  -o PATH.TXT           Output .txt file for path indices
  --out-z Z.PATH.TXT    Output .txt file for path z-values

```

--------------------------------

### CryoDRGN Downsample Help

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Displays the help message for the `cryodrgn downsample` command, outlining all available options and arguments.

```bash
(cryodrgn) $ cryodrgn downsample -h
usage: cryodrgn downsample [-h] -D D -o MRCS [-b B] [--is-vol] [--chunk CHUNK] [--datadir DATADIR]
                           [--max-threads MAX_THREADS] [--ind PKL]
                           mrcs

Downsample an image stack or volume by clipping fourier frequencies

positional arguments:
  mrcs                  Input particles or volume (.mrc, .mrcs, .star, .cs, or .txt)

options:
  -h, --help            show this help message and exit
  -D D                  New box size in pixels, must be even
  -o MRCS               Output projection stack (.mrcs)
  -b B                  Batch size for processing images (default: 5000)
  --is-vol              Flag if input .mrc is a volume
  --chunk CHUNK         Chunksize (in # of images) to split particle stack when saving
  --datadir DATADIR     Optionally provide path to input .mrcs if loading from a .star or .cs file
  --max-threads MAX_THREADS
                        Maximum number of CPU cores for parallelization (default: 16)
  --ind PKL             Filter particle stack by these indices

```

--------------------------------

### CLI Command: cryodrgn abinit

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-ai-ab-initio-reconstruction/cryodrgn-abinit-command-api

Executes the abinit reconstruction process with required particle data and configuration parameters.

```APIDOC
## cryodrgn abinit

### Description
Performs ab initio 3D reconstruction from a particle stack using the cryoDRGN framework.

### Parameters
#### Required Parameters
- **particles** (string) - Required - The input particle stack (.mrcs, .star, .cs, or .txt).
- **--outdir/-o** (string) - Required - The directory where model outputs will be saved.
- **--ctf** (string) - Required - Path to CTF parameters (.pkl). Must be the size of the full dataset.
- **--zdim** (int) - Required - Dimension of conformation z-latent-space (0 for homogeneous reconstruction).

#### Other Input Parameters
- **--ind** (string) - Optional - Path to indices (.pkl) or number of images to keep.
- **--datadir** (string) - Optional - Path to the directory containing the .mrcs files.
- **--relion31** (flag) - Optional - Flag for RELION3.1 .star file format.
- **--uninvert-data** (flag) - Optional - Flag for not inverting input data.
- **--load** (string) - Optional - Path to a checkpoint to load.
- **--load-poses** (string) - Optional - Path to a pose file to load.
- **--lazy** (flag) - Optional - Flag for lazy data loading.
- **--max-threads** (int) - Optional - Number of threads to use for data loading.
- **--no-trans** (flag) - Optional - Indicates dataset does not contain translations.
- **--norm** (list) - Optional - Override default normalization (mean, std).

#### Training Scheduling
- **--num-epochs** (int) - Optional - Total epochs to train.
- **--epochs-sgd** (int) - Optional - Epochs to train for SGD.
- **--epochs-pose-search** (int) - Optional - Epochs for hierarchical pose search.
- **--n-imgs-pose-search** (int) - Optional - Number of images to use for hierarchical pose search.

#### Loss
- **--beta-conf** (float) - Optional - Beta term penalizing KL divergence (default: 0.0).
- **--trans-l1-regularizer** (float) - Optional - Strength of L1 regularizer on translations (default: 0.0).
- **--l2-smoothness-regularizer** (float) - Optional - Strength of L2 smoothness regularization (default: 0.0).

#### Optimizers
- **--hypervolume-optimizer-type** (string) - Optional - Optimizer for hypervolume (default: "adam").
- **--pose-table-optimizer-type** (string) - Optional - Optimizer for pose table (default: "adam").
- **--conf-table-optimizer-type** (string) - Optional - Optimizer for conformation table (default: "adam").
- **--conf-encoder-optimizer-type** (string) - Optional - Optimizer for conformation encoder (default: "adam").
- **--lr** (float) - Optional - Learning rate of hypervolume (default: 1e-4).
- **--lr-pose_table** (float) - Optional - Learning rate of pose table (default: 1e-3).
- **--lr-conf-table** (float) - Optional - Learning rate of conformation table (default: 1e-2).
- **--lr-conf-encoder** (float) - Optional - Learning rate of conformation encoder (default: 1e-4).
- **--wd** (float) - Optional - Weight decay (default: 0.0).

#### Masking
- **--output-mask** (string) - Optional - Type of output mask (default: "circ").
- **--add-one-frequency-every** (int) - Optional - Frequency for adding new frequencies (default: 100,000).
- **--n-frequencies-per-epoch** (int) - Optional - Frequencies to add per epoch (default: 10).
- **--max-freq** (int) - Optional - Highest frequency to use.
- **--window-radius-gt-real** (float) - Optional - Radius of circular mask (default: 0.85).
- **--l-start-fm** (int) - Optional - Starting size for frequency marching (default: 12).

#### Conformations
- **--variational-het** (bool) - Optional - Activate variational mode (default: False).
- **--std-z-init** (float) - Optional - Std dev of initial conformations (default: 0.1).
- **--use-conf-encoder** (bool) - Optional - Use encoder to predict conformations (default: False).
- **--depth-cnn** (int) - Optional - Depth of encoder (default: 5).
- **--channels-cnn** (int) - Optional - Channels in encoder (default: 32).
- **--kernel-size-cnn** (int) - Optional - Kernel size in encoder (default: 3).
- **--resolution-encoder** (int) - Optional - Resolution of images for encoder.
```

--------------------------------

### Show Help for parse_ctf_csparc

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-empiar-10076-tutorial

Displays the usage and available options for the `cryodrgn parse_ctf_csparc` command, which parses CTF parameters from a cryoSPARC particles.cs file.

```bash
(cryodrgn) $ cryodrgn parse_ctf_csparc -h
usage: cryodrgn parse_ctf_csparc [-h] -o O [--png PNG] [-D D] [--Apix APIX] cs

Parse CTF parameters from a cryoSPARC particles.cs file

positional arguments:
  cs           Input cryosparc particles.cs file

optional arguments:
  -h, --help   show this help message and exit
  -o O         Output pkl of CTF parameters
  --png PNG    Optionally plot the CTF

Optionally provide missing image parameters:
  -D D         Image size in pixels
  --Apix APIX  Angstroms per pixel
```

--------------------------------

### Restart and Extend cryoDRGN-AI Training

Source: https://ez-lab.gitbook.io/cryodrgn/cryodrgn-ai-ab-initio-reconstruction/cryodrgn-ai-ab-initio-empiar-10076-tutorial

Use the --load argument with a saved checkpoint to resume interrupted training or extend training for more epochs. Specify a new value for --num-epochs to set the total number of epochs.

```bash
cryodrgn abinit particles.128.mrcs --zdim 4 --lazy --ctf ctf.pkl --load 50S_abinit/001_defaults.128/weights.25.pkl -n 40 -o 50S_abinit/001_defaults.128/ --uninvert-data
```
Tokens:
Raw