Experiments
Task types
Ab-initio Heterogeneous Reconstruction
Description: Reconstruct a single (homogeneous) or multiple (heterogeneous) 3-D maps from a set of particles, without any initial models or starting structures required. See: How to perform Ab-Initio Reconstruction in cryoSPARC.
Notes: If no initial map is provided, it is possible to discover 3-D classes that are significantly different. If an initial map is provided, it will be used as the starting structure for all classes.
Limitations: Ab-initio reconstruction can fail when data quality is poor or when viewing directions are missing or strongly biased. Highly symmetric structures can also pose challenges. It is not recommended to enforce symmetry during ab-initio reconstruction but can be helpful when symmetry is known in advance.
Input:
- Initial model (optional)
- Particle meta data (optional; used for selecting a class from a previous experiment)
- Parameters (generally no changes necessary to defaults, except the number of classes)
Output:
- 3-D Maps (one or more)
- Particle meta data (contains alignments and classifications)
- Plots, including orientation distributions
[NEW] 2D Classification
Description: Classify particles into multiple 2D classes to facilitate stack cleaning and removal of junk particles. Also useful as a sanity check to investigate particle quality. See: How to do 2D Classification in cryoSPARC.
Notes: Should be used in conjunction with 3D classification to purify particle sets.
Input:
- Particle meta data (optional; used for selecting a class from a previous experiment)
- Parameters (generally no changes necessary to defaults, except the number of classes)
Output:
- 2D class averages
- Particle meta data (contains alignments and classifications)
- Plots
[NEW] Select 2D Classes
Description:
Notes:
Input:
-
Output:
-
[NEW] Union/Intersect Particles
Description:
Notes:
Input:
-
Output:
-
[NEW] Align 3D Maps
Description: See: How to use Align 3D Maps
Notes:
Input:
-
Output:
-
Homogeneous Refinement
Description: Rapidly refine a single homogeneous structure to high-resolution (measured by gold-standard FSC). See: How to Refine structures to high-resolution in cryoSPARC.
Notes: This job is generally used to refine maps that were output from ab-initio reconstruction, using new optimized codepaths and GPU kernels. By default, this experiment uses dynamic masking to automatically generate a soft mask used during refinement. Thresholds and distances for this dynamic mask can be set, or a user-provided mask can be specified instead, if necessary.
Limitations: The initial model provided must be on the correct grey scale. Outputs from ab-initio reconstruction in cryoSPARC meet this requirement.
Input:
- Initial model
- Particle meta data (optional; used for selecting a class from a previous experiment)
- Parameters (generally no changes necessary to defaults)
Output:
- Refined 3-D map
- Half-maps
- Mask used in refinement
- Mask used in FSC calculation
- Gold-standard FSC curve
- Plots, including orientation distributions
[NEW] Heterogeneous Refinement (3D Classification)
Description: Rapidly classify and refine _n_ structures starting with _n_ initial models or references, generated in cryoSPARC or elsewhere. See: How to do Heterogeneous Refinement in cryoSPARC.
Notes: This task enables simultaneously sorting particles and identifying classes and is particularly useful in cases where identified classes look very similar. This task can also be used as a method to remove “junk” particles.
Limitations: The initial models provided must be on the correct grey scale. Outputs from ab-initio reconstruction in cryoSPARC meet this requirement.
Input:
- Initial model
- Particle meta data (optional; used for selecting a class from a previous experiment)
- Parameters
Output:
- Refined 3-D map
- Half-maps
- Mask used in refinement
- Mask used in FSC calculation
- Plots, including orientation distributions
- Particle alignments and assignments into classes
[NEW] ResLog Analysis
Description: Obtain a ResLog plot to better understand how resolution changes as you add more particles to a refinement. See: How to perform ResLog Analysis
Notes: Can be used after any refinement experiment to determine how good the result is and to tell whether a 3D classification was useful.
Limitations: Currently only supports “plain” ResLog plots, not parameter or defocus-based plots.
Input:
- Particle meta data from a refinement experiment
- Parameters
Output:
- ResLog plot
[NEW] Generate Mask
Description: Generate a mask by thresholding a density, then dilating and softening. Identify a region of interest following refinement and mask out unstructured or irrelevant regions to enhance the resolution of the region of interest in a further refinement using the mask. See: How to use Generate Mask
Notes: The Threshold Value are on the same greyscale as the input density and should correspond to threshold values from Chimera or other viewer.
Input:
- 3-D map
Output:
- 3-D mask
Sharpen, Resample, Align, Flip
Description: Post-process maps for visualization.
Notes: Output maps from refinement are not always optimally sharpened for visualization.
Resampling a result map to a higher box-size can improve the smoothness of meshes generated in Chimera, especially for very high resolution results.
Ab-initio reconstruction will often reconstruct maps with incorrect handedness, so they need to be flipped.
Limitations: There is a known bug with very high b-factors when sharpening.
Input:
- 3-D map
Output:
- 3-D map
[NEW] Disk Cleanup
Description: Disk cleanup utility that enables user to remove experiments that are no longer accessible by the UI. Also allows user to select experiments belonging to a particular dataset and either completely erase them or just delete their intermediate results to minimize disk usage.
Notes: Only admin users are able to delete experiments created by other users, and are able to delete experiments from multiple datasets at once.
Import Volume / Mask
Description: Import a 3D volume or mask into cryoSPARC from another source.
Notes: MRC files are supported and can be uploaded directly through the web interface. When importing a mask, ensure to set the “Imported Volume is a mask” parameter to true.
Input:
- 3-D map or mask
Launch a new experiment
For sample experiments, please view our case studies on the 80S ribosome and the T20S proteasome.
1 From a selected dataset
From the ‘Datasets’ page, click to select the dataset.
Click on the ‘Experiments’ page, then click on ‘New Experiment’. The drop down menu displays the experiments applicable/available to/for the chosen dataset in purple. To view more information about the experiment type, hover over the ‘?’ symbol next to the experiment name:
When starting an experiment in cryoSPARC, structure and particle set inputs are selected directly from the experiments page, directly below the list of applicable experiments. For ab-initio reconstruction, no initial structures need to be selected, but the number of ab-initio classes to attempt reconstruction for must be specified. The default number of ab-initio classes is set to 1. The value need not be changed if performing a homogeneous (i.e., single class) ab-initio reconstruction. For heterogeneous (i.e., multiple class) reconstruction, you can input the desired number of ab-initio classes.
Click on the desired experiment type, e.g., ‘Ab-initio’, which will take you to the ‘Details’ page. This provides a place for making notes about the experiment. To change the name of the experiment from the default ‘New Experiment’, simply click on the heading and modify the text.
Optional: To adjust experiment parameters, navigate to the ‘Setup’ tab on the left hand sidebar. CryoSPARC provides default parameters for each task type, and no modifications need to be made to achieve the reported results.
Click ‘Launch’ and ‘Enqueue’ to commence the experiment and view/download plots and results in real-time. cryoSPARC’s job scheduler will automatically run queued jobs as computational resources (GPUs, RAM) become available. CryoSPARC can be configured to only run jobs on specific GPUs, using the
cryosparc configurecommand.
CryoSPARC uses purple status indicators to show that an experiment is running. Blue is used to indicate a queued experiment, green indicates a completed experiment, and grey/blank indicates that an experiment has not yet been started. Orange indicates that a job was killed. CryoSPARC jobs will continue to run even if you close your browser.
Tips and examples:
- Ab-initio reconstruction workflow
- Discover previously unknown conformations using ab-initio reconstruction
- Refinment workflow
- Sub-classification workflow
- 2D Classification: how to
- Heterogeneous Refinement (3D Classification): how to
- ResLog Analysis, Generate Mask and Align 3D Maps: how to
Troubleshooting:
- Ab initio reconstruction fails to start, error in mrc.py
- No heartbeat error
- cryoSPARC on OpenSuse 13
2 From the result of a previous experiment
Navigate to the ‘Experiments’ page and locate a completed experiment from which you would like to run a new experiment. Note: You can download the result structure of the previous experiment by clicking on the double chevron on the image projection.
Click ‘New Experiment’ to display the available experiment types.
The ‘Structure’ and ‘Particles’ buttons below the image projection can be clicked once or multiple times to select the number of result structures and particles you wish you include in the new experiment. Note: The list of available experiments will change to reflect the combination of structure and particles you choose, and the ‘cart’ below the list of available experiments will reflect your choices.
Select the experiment type, which will take you to the ‘Details’ page.
Optional: To adjust experiment parameters, navigate to the ‘Setup’ tab on the left hand sidebar. CryoSPARC provides default parameters for each task type, and no modifications need to be made to achieve the reported results.
Click ‘Launch’ and ‘Enqueue’ to commence the experiment and view/download plots and results in real-time.
Modify or stop an experiment
To modify the parameters and re-start an experiment, click ‘Clear’ and then change the parameters by clicking ‘Setup’ on the left-hand sidebar. To re-start the experiment, click ‘Launch’ and ‘Enqueue.
To stop an experiment, click ‘Kill’. An experiment that has been killed cannot be resumed from the kill point.
View and download experiment progress in real-time
When the job begins to run, results will stream in real-time on the Launch page. These include plots and result files that are generated. For example, three axis-aligned slices of each ab-initio structure are created in the results stream that can be useful for inspecting the progress of reconstruction:
Several other useful plots are provided in-line as well. We downloaded the below viewing direction plot in PNG format directly from the results stream. PDF and raw data formats are also available for making publication figures. The below plot shows a typical viewing direction distribution over azimuth and elevation angles for this dataset:
Select ‘Following latest to see the most recent events or ‘Show from top’ to view all events in chronological order.
Download final result
When an experiment is complete, the result structure and result meta data file can be downloaded directly from the Launch page. Alternatively, the file path(s) can be copied from the Launch page.
As mentioned above, you can also download the result structure from the ‘Experiments’ page below the image projection.
Tips:
Interpreting results
Checking for preferred orientation
The Viewing Direction Distribution plot, available at the completion of a Refinement experiment, sheds light on the degree of orientation bias in your sample. For example, the following plot shows that while there are some viewing directions from which the sample was never imaged, there are images available from nearly every viewing direction for the sample, and the distribution of those viewing directions does not indicate a large amount of orientation bias.
On the other hand, in the below plot, it is clear that nearly all of the top and bottom viewing directions are missing in the sample, i.e., we almost only ever see the sample from the sides, not from the top or bottom. However, the band of equatorial views is likely wide enough that we are able to see enough of the sample from a variety of views to generate a decent reconstruction and refinement.
Finally, the Viewing Direction Distribution plot will reflect the symmetric nature of a dataset containing a highly symmetric protein - e.g., the below plot shows a structure with four-fold symmetry.
Interpreting gold-standard FSC curves
No Mask:
This is the raw FSC calculated between two independent half-maps reconstructed from the data. There is no masking applied, so both the structure and solvent are included in this FSC.
Spherical:
This is the FSC calculated after applying a soft spherical mask to both half maps. The outer radius of the soft sphere is equal to half the volume box-size (i.e. the sphere extends to the faces of the box in all directions). The inner radius is 85 percent of the outer radius. Between inner and outer radii, a soft cosine edge transitions from a mask value of one to a value of zero.
Loose:
This is the FSC calculated after applying a soft solvent mask to both half maps. The loose mask is calculated as follows. First, the density map is thresholded at 50% of the maximum density value. The resulting volume is dilated to create a soft mask. Voxels in the mask that are within 25 angstroms of the thresholded region receive a mask value of 1.0. Voxels between 25 and 40 angstroms fall off with a soft cosine edge, and voxels outside 40 angstroms receive a value of 0.0.
Tight:
This is the same as the loose mask, except the dilation distances are 6 angstroms for the value 1.0 distance and 12 angstroms for the value 0.0 distance.
Corrected:
This is the FSC curve calculated using the tight mask with correction by noise substitution [1]. The two half maps have their phases randomized beyond a certain resolution, then the tight mask is applied to both, and an FSC is calculated. This FSC is used along with the original FSC before phase randomization to compute the corrected FSC as in [1]. This accounts for correlation effects induced by masking. The resolution at which phase randomization begins is the resolution at which the no-mask FSC drops below the FSC = 0.143 criterion.
[1] Chen, S. et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy 135, 24–35 (2013).