|
|
| Line 1: |
Line 1: |
| CryoNAV's processing pipeline is built around IMOD's core utilities and extended with deep-learning denoising methods. The pipeline is accessed through two complementary approaches:
| | #REDIRECT [[CryoNAV Overview Tomogram Processing]] |
| | |
| * '''Per-tilt-series''' -- the interface mirrors the step-by-step workflow familiar from IMOD's eTomo: navigate to a tilt series, run a processing step, review results (thumbnails, alignment residuals, CTF plots), and proceed to the next step with adjusted parameters if needed.
| |
| * '''Batch''' -- select a batch of tilt series and submit them all for processing with a single action, applying the same parameters across the set.
| |
| | |
| == Pipeline steps ==
| |
| | |
| The processing pipeline for a typical tilt series:
| |
| | |
| * '''Motion correction''' -- correcting beam-induced specimen movement within individual tilt images.
| |
| * '''CTF estimation and correction''' -- measuring and fitting the contrast transfer function for defocus determination.
| |
| * '''Tilt series alignment''' -- aligning the tilt series using fiducial markers or patch tracking.
| |
| * '''Tomographic reconstruction''' -- computing the 3D volume via weighted back-projection or SIRT.
| |
| * '''Tomogram denoising''' (CryoCARE, DeepDeWedge) -- deep-learning denoising to improve contrast and signal-to-noise ratio. Denoising can also be applied to facility-imported data, requiring only the raw frames and reconstruction parameters.
| |
| | |
| == Job submission ==
| |
| | |
| At submission time, the user selects one or more tilt series, chooses a processing template and a computing template, and optionally adjusts parameters via a visual form. CryoNAV constructs the appropriate commands and submits them to the selected execution backend.
| |
| | |
| Processing steps can be chained into multi-step workflow templates with automatic dependency tracking. A standard workflow might be: motion correction -> alignment + CTF estimation -> reconstruction -> optional denoising. Each step is submitted once its predecessor completes successfully; failed steps halt the remainder of the chain.
| |
| | |
| == Local and HPC execution ==
| |
| | |
| CryoNAV supports both local workstation execution and HPC cluster submission (via SLURM) within the same deployment, and both modes are available simultaneously: a user might run a quick test alignment locally before submitting a full batch to the cluster. The execution backend is abstracted behind a common interface, so job submission, progress monitoring, log viewing, and error handling work identically regardless of where the job runs.
| |
| | |
| == Progress monitoring ==
| |
| | |
| CryoNAV provides real-time progress tracking for running jobs through two complementary strategies, adapted from the approach used by eTomo:
| |
| | |
| * '''Log parsing''' -- processing commands (both IMOD utilities and denoising tools) that print progress messages to stdout (e.g., "Processing frame 15/60" from alignframes) are parsed with regular expressions to extract the current step and total steps, yielding a percentage completion.
| |
| * '''File size monitoring''' -- for steps that produce output files of predictable size, CryoNAV monitors intermediate file growth and compares it to the expected final size derived from input metadata.
| |
| | |
| The web interface displays progress bars with smooth interpolation between server updates and human-readable ETA estimates.
| |
| | |
| == Immutable processing branches ==
| |
| | |
| Each processing run is stored as an immutable record. Re-running a step with different parameters creates a new parallel branch rather than overwriting the original, allowing side-by-side comparison of parameter choices. See [[Key Concepts#Immutable processing branches|Immutable processing branches]] for the full discussion.
| |
| | |
| == See also ==
| |
| | |
| * [[Key Concepts]]
| |
| * [[Integration with Cryo-EM Tools]]
| |
| * [[Tilt Series Data Import]]
| |
| | |
| [[Category:Processing]]
| |
