Cryo-ET / Tomography¶

RECOVAR supports tilt-series data for cryo-ET heterogeneity analysis, including the use of a focus mask.

Experimental

Cryo-ET support is newer than SPA support and may be less stable. No paper has been published on this feature yet.

RELION-5 input only

RECOVAR's cryo-ET support currently reads RELION-5 tilt-series data. Process your tomograms through the RELION-5 tomography pipeline (import → tilt-series alignment → particle extraction / Bayesian polishing) to produce the tomograms.star and particles.star that RECOVAR consumes. See the RELION-5 tomography tutorial for how to generate those files.

Using the GUI¶

In the Web GUI cryo-ET is a single checkbox. In the Pipeline form, point Particles at a RELION5 2D-tilt star (produced by recovar parse_relion5_tomo, see below), expand Advanced, and tick Tilt series. A cryo-ET panel appears -- per-tilt dose, tilt angles, and the CTF model are read automatically from the star, so the only common knob is Max tilts (how many tilts per series to use).

Pipeline form with the Tilt series option enabled

The rest of this page covers the same workflow from the command line.

Importing from RELION5¶

RECOVAR works on a 2D-tilt star (one row per particle per tilt). Starting from the RELION5 outputs — a tomograms.star with the tilt-series geometry and a particles.star with the 3D particle positions/orientations — you can get there two ways.

Let the pipeline convert (recommended for the CLI)¶

Pass the RELION5 particles.star as the input and point --tomograms at the geometry star. The pipeline runs the conversion internally before processing, and --tomograms implies --tilt-series:

recovar pipeline Extract/job260/particles.star -o output --mask mask.mrc \
    --tomograms Polish/job249/tomograms.star

Convert explicitly with `parse_relion5_tomo`¶

To materialize the 2D-tilt star yourself — for example to load it in the GUI, which takes the 2D star directly — run the conversion as its own step:

recovar parse_relion5_tomo \
    -t Polish/job249/tomograms.star \
    -p Extract/job260/particles.star \
    -o particles_2d.star

This reads the RELION5 3D tomography metadata and produces a 2D STAR file where each row is one tilt of one particle, with per-tilt defocus, orientation, and dose information. The output is directly compatible with recovar pipeline --tilt-series.

Requirements:

tomograms.star from a Polish or Tomograms job (contains tilt-series geometry)
particles.star from an Extract or Refine job (contains 3D particle positions and orientations)

Tilt image dimensions are auto-detected from the MRC headers.

Credits

Projection geometry adapted from relion2cryodrgn by Ryan Feathers (Princeton/Zhong lab), based on code by Bogdan Toader (MRC-LMB/RELION team).

Usage¶

recovar pipeline particles_2d.star -o output \
    --mask mask.mrc --tilt-series

The input is a 2D STAR file with tilt-series metadata (one row per particle per tilt, grouped by _rlnGroupName).

Options¶

Flag	Default	Description
`--tilt-series`	False	Enable tilt-series mode
`--tilt-series-ctf`	Auto	CTF model: `cryoem`, `relion5`, `warp`
`--dose-per-tilt`	From file	Dose per tilt in e/A^2
`--angle-per-tilt`	From file	Tilt angle increment
`--ntilts`	All	Maximum number of tilts to use

CTF models¶

Model	Description
`cryoem`	Standard cryo-EM CTF (for subtomogram averaging)
`relion5`	RELION 5 tilt-series CTF with dose weighting
`warp`	Warp-style CTF

The default is relion5 for tilt-series data and cryoem otherwise. The default is correct for RELION5 input, so the GUI does not expose it. warp is experimental.

With focus mask¶

Add a focus mask with --focus-mask:

recovar pipeline particles.star -o output \
    --mask mask.mrc --focus-mask binding_site.mrc --tilt-series

Tips¶

For cryo-ET data, the --maskrad-fraction default (20) may need adjustment. Lower-resolution data may benefit from increasing this value.
The --n-min-particles default (100) may need to be reduced for smaller tomography datasets.
Use --ntilts to limit the number of tilts if some have poor quality.