What's new in v2.8

May 30, 2019

Here's an overview of the latest updates to cryoSPARC:

Migrating a cryoSPARC Instance

New

Our latest tutorial covers how to move your cryoSPARC instance from one location to another - either between folders, different network storage locations, or even different host machines entirely:

Local Patch-Based Motion Correction Beta

New

A very fast auto-tuning patch-based local motion correction method, built on top of the original cryoSPARC Local Motion correction method (which is a variant of alignparts). On modern GPU/CPU/disk hardware, less than 10 seconds per movie (single GPU) for global full-frame + local anisotropic motion estimation and correction. High-quality interpolation. No need for particle pick locations. Outputs both non-dose-weighted and dose-weighted micrographs, ready for CTF estimation and extraction. Multi-GPU mode enabled for blazing fast preprocessing speed. Native support for parallelized TIFF and MRC.BZ2 decompression.

Local Patch-Based CTF Estimation Beta

New

A new fast auto-tuning patch-based local CTF estimation method. Deals easily with tilt data, bent ice, very small particles, wide range of defocus, phase plate data. Fits a locally-variable CTF landscape to each micrograph. No prior knowledge about tilt/etc needed. Based on a few new robust methods for reference-free background and CTF envelope estimation, and LBFGS optimization to maximize simulated 2D CTF fit. Outputs per-particle local CTF estimates, in ~3 seconds per micrograph (single GPU). Multi-GPU mode enabled for blazing fast preprocessing speed.

Create Templates

New

Create 2D templates for template-picking from any 3D volume. Useful for iterating particle picking after obtaining an ab-initio 3D structure or refinement, especially when rare views are difficult to pick without good templates.

Blob Picker

New

Fully automatic fast blob-based particle picker. Can use both circular and elliptical blobs, to easily pick particles of various shapes and elongation. Less than 1 second per micrograph (single GPU).

Other Improvements

New

Support for movies produced by K3 Camera (including fixes for quirks of Leginon/Gatan Latitude)

  • Enhancement Memory usage for motion correction and picking is now minimized for super-res K3 movies
  • Fixed Particles extracted from K3 Movies produced NaN values
  • Fixed MRC mode 0 is now interpreted as uint8 rather than int8, to match Latitude for K3 movies
New

heartbeat time is now configurable (the amount of time that cryoSPARC core waits before labeling a job as failed if there is no response from the parent process running the job). Addresses problem where long-running jobs stop communicating with the master process intermittently due to network issues.

  1. Edit the file cryosparc2_master/config.sh and append export CRYOSPARC_HEARTBEAT_SECONDS="360" to the end
  2. Restart cryoSPARC: cryosparcm restart
New
Network requests between worker and master processes now by default have a 300 second timeout (instead of 120s) and also retry 3 times in case the host is unreachable. This should reduce the occurrence of failed jobs due to network timeout issues or slowdowns of the master process
New
Ability to mark failed or killed jobs in cryoSPARC as completed allowing it's latest outputs to be used for further processing) - use the new "Mark Job as Complete" button in the Job Details panel
New
Import .star files now also imports rlnRandomSubset to preserve the gold-standard split and rlnAngle* and rlnOriginX/Y to preserve alignments of particles
New
Ab-initio reconstruction now has a parameter to limit it to running on only a subset of particles, to more easily deal with very large (million+ particle) datasets. You can use the volumes generated by a 'partial' ab-initio reconstruction as the initial references for a heterogeneous refinement, which will then quickly classify and refine the rest of the dataset
Update
Box size range in interactive jobs increased to [4, 1024]
Update
MRC Extended header size (NSYMBT) taken into account when reading MRC files
Fixed
Power threshold slider in interactive jobs can be more finely controlled
Fixed
NCC and Power Thresholds for picking are limited to 1st and 99th percentile to reduce number of outliers
Fixed
Bug in GPU extraction that would cause job to hang indefinitely after starting
Fixed
Bug in dynamic mask generation in Local Refinement when Non-uniform Refinement is turned on)
Fixed
Off-by-one bug in fulcrum selection in Local Refinement


Stay Informed

We are always working on new features and updates! Subscribe to the cryoSPARC newsletter to stay on top of new features, updates and best practices for data processing, or follow us on Twitter.

This website uses cookies to ensure you get the best experience. To learn more, please refer to our Privacy Policy