Written and maintained by CASRAI Editorial Board
Last updated
CryoSPARC is software that turns raw cryo-electron microscopy (cryo-EM) images into a three-dimensional density map of a protein or other macromolecule. This guide is for structural biologists, core-facility staff and research computing teams who need to know what it does, how to obtain and install it, what a first processing run looks like, and where the results go afterwards. Facts below come from the CryoSPARC guide at guide.cryosparc.com, the EMBL-EBI EMDB site, the RELION documentation and the peer-reviewed literature, read on 2026-11-07.
What is CryoSPARC and what is it used for in research?
CryoSPARC is a software package for processing single-particle cryo-EM data. It was introduced by Punjani, Rubinstein, Fleet and Brubaker in Nature Methods 14(3):290-296 (2017, doi:10.1038/nmeth.4169). The paper describes an approach that combines stochastic gradient descent with branch-and-bound maximum-likelihood optimisation and allows ab initio reconstruction, meaning a first 3D model can be built from the images without a prior structure. That 2017 paper is the reference to cite when you report use of the software.
In practice the program takes movies from an electron microscope, corrects beam-induced motion, estimates the contrast transfer function, picks and extracts particles, sorts them into 2D and 3D classes, and refines a final map. It runs as a web application on a master process, with GPU worker processes doing the computation.
What is a cryo-EM map?
A cryo-EM map is the three-dimensional density volume reconstructed from many two-dimensional images of frozen particles. It is the main result of a CryoSPARC refinement job. The EMDB (Electron Microscopy Data Bank, run by EMBL-EBI) describes itself as a public repository for cryogenic-sample electron microscopy volumes and representative tomograms of macromolecular complexes and subcellular structures, and lists single-particle analysis, helical reconstruction, electron tomography, subtomogram averaging and electron crystallography among the techniques it holds. It listed 64,153 entries as of 4 November 2026. Raw image data can be deposited separately: see EMPIAR, the public image archive, and, for the atomic models built into maps, the Protein Data Bank.
What are the disadvantages of cryo-EM?
The software is rarely the first limit; sample preparation usually is. A 2018 review in Acta Crystallographica Section D (Drulyte et al., 74(6):560-571, doi:10.1107/S2059798318006496) states that for many single-particle projects major bottlenecks remain at the sample-preparation stage. It lists these specific problems:
- obtaining grids of sufficient quality can require careful optimisation of many variables;
- problems with the sample itself, such as buffer components and labile complexes;
- sample distribution, including concentration and affinity for the support film;
- preferred orientation of particles;
- poor reproducibility of grid making within and between batches.
For how specimen preparation limits attainable resolution in electron microscopy more generally, see transmission electron microscopy sample preparation and resolution. For a method that works from crystals instead of frozen particles, see X-ray crystallography.
How to access CryoSPARC (official source only)
According to the CryoSPARC guide, the software is free for non-profit academic use. You request a License ID through the form on the download page at cryosparc.com/download. For-profit use is licensed separately through Structura Biotechnology (contact address given in the guide: [email protected]). Many institutions run CryoSPARC on a shared cluster, in which case access is a matter of asking your facility for an account on its existing instance rather than installing it yourself.
How to install CryoSPARC
Requirements stated in the guide
- At least 15 GB of disk for the packages.
- An installation path of no more than 83 characters.
- Worker nodes need an NVIDIA GPU with driver 570.26 or newer (x86-64) or 580.65.06 or newer (ARM64); the master runs on x86-64 only.
- The master’s ports must not be directly reachable from the internet, and the filesystem holding the database must not fill up.
- The
--cudapathoption applies only to versions before v4.4.0.
Download
Set LICENSE_ID and VERSION, then:
curl -L "https://get.cryosparc.com/download/master-$VERSION/$LICENSE_ID" -o cryosparc_master.tar.gz
curl -L "https://get.cryosparc.com/download/worker-$VERSION/$LICENSE_ID" -o cryosparc_worker.tar.gz
tar -xf cryosparc_master.tar.gz cryosparc_master
Single workstation (master and worker together)
cd cryosparc_master
./install.sh --standalone --license $LICENSE_ID --worker_path <worker path> --ssdpath <ssd path> --initial_email <email> --initial_username "<user>" --initial_firstname "<given>" --initial_lastname "<surname>"
Optional flags include --port <n> and --initial_password <pw>. Then open http://<workstation_hostname>:<base_port_number> in a browser and sign in as the admin user you created. The guide’s examples use port 61000 for the master.
Separate master and worker (cluster setups)
# master
./install.sh --license $LICENSE_ID --hostname <master_hostname> --dbpath <db_path> --port <port_number>
./bin/cryosparcm start
cryosparcm user create --email ... --password ... --username ... --firstname ... --lastname ... --role admin
# worker
cd cryosparc_worker
./install.sh --license $LICENSE_ID
./bin/cryosparcw connect --master <master_hostname> --worker <worker_hostname> --port <port_num>
For a scheduler-managed cluster the guide uses cryosparcm cluster connect with a cluster_info.json and a cluster_script.sh. Other install flags (--insecure, --allowroot, --yes) exist; the guide’s installation page lists them for each version.
A first workflow: the introductory tutorial, step by step
CryoSPARC’s own introductory tutorial (v4.0 and later) processes a 20-movie subset of EMPIAR-10025, the T20S proteasome. Get the data on the master with cryosparcm downloadtest and extract it with tar xvf. In the web interface, press New Project, create a workspace, then build jobs in this order. Import values for this dataset are pixel size 0.6575 Å, 300 kV, Cs 2.7 mm, total dose 53 e/Ų.
- Import Movies
- Patch Motion Correction
- Patch CTF Estimation
- Micrograph Denoiser (v4.5 and later, optional)
- Blob Picker (particle diameter 100-200 Å)
- Micrograph Junk Detector (v4.7 and later, optional) and Inspect Particle Picks
- Extract From Micrographs (box size 440; about 11,129 particles)
- 2D Classification (50 classes), then Select 2D Classes
- Template Picker (diameter 190 Å), then Inspect, Extract, 2D Classification and Select 2D Classes again
- Ab-Initio Reconstruction
- Homogeneous Refinement (D7 symmetry for this specimen)
- Volume Tools / masking and Validation (FSC); Sharpening Tools is optional
The Tree View (flowchart icon) shows how jobs connect. Job names and version tags above are from the current guide; menu items can change between releases, so follow the guide for the version you installed.
Output and how to read it
| Stage | Job | What comes out |
|---|---|---|
| Preprocessing | Patch Motion Correction, Patch CTF Estimation | Motion-corrected micrographs and per-micrograph CTF estimates |
| Picking | Blob Picker, Template Picker, Inspect Particle Picks | Particle picks |
| Extraction | Extract From Micrographs | Boxed particle images |
| 2D | 2D Classification, Select 2D Classes | Class averages and the subset of particles you keep |
| 3D | Ab-Initio Reconstruction, Homogeneous Refinement | Initial volume, then a refined map (map_sharp) with gold-standard FSC (GSFSC) plots |
The final map_sharp volume can be downloaded from the refinement job’s Outputs tab. Resolution is read from the FSC curves. A map intended for publication is then deposited, with the model, as described in the EMDB and PDB guides linked above.
Troubleshooting installation
- Installer fails or paths break: check that the install path is 83 characters or fewer.
- GPU not usable by the worker: compare the NVIDIA driver with the minimum above (570.26 on x86-64, 580.65.06 on ARM64).
- Browser cannot reach the interface: confirm the hostname and base port you passed to the installer, and that the master was started with
cryosparcm start. - Installing as root: the installer needs the
--allowrootflag. - Database problems: the guide warns that the database filesystem must not fill up.
Free alternatives
| Software | What the official documentation says |
|---|---|
| RELION | Cryo-EM structure determination software using an empirical Bayesian approach, developed in Sjors Scheres’s group at the MRC Laboratory of Molecular Biology. Distributed under a GNU General Public License; the documentation says anyone, including commercial users, can download, use and modify it for free. |
The RELION documentation asks users to cite the project’s papers; see its publications page for the reference that matches the version you used.
Frequently asked questions
What is CryoSPARC?
Software for processing single-particle cryo-EM data into 3D maps, described in Punjani et al., Nature Methods 2017. It is free for non-profit academic use under a licence requested from the vendor.
How do I access CryoSPARC?
Request a License ID at cryosparc.com/download for academic use and install it with the commands above, or ask your institution whether it already runs a shared instance. For-profit licensing goes through Structura Biotechnology.
What is a cryo-EM map?
The 3D density volume reconstructed from particle images; public maps are archived in EMDB.
What are the disadvantages of cryo-EM?
Sample and grid preparation is a common bottleneck, including preferred orientation and poor reproducibility between grid batches.
Is there a public CryoSPARC login page?
No single one. Each installation has its own web address (the hostname and base port set at install), and a facility may publish its own.
Sources (read 2026-11-07)
- CryoSPARC guide: How to download, install and configure; Introductory tutorial.
- Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. cryoSPARC. Nat Methods 14(3):290-296, 2017. PubMed 28165473.
- EMBL-EBI, Electron Microscopy Data Bank.
- Drulyte I et al. Acta Cryst D 74(6):560-571, 2018. doi:10.1107/S2059798318006496.
- RELION documentation.







