cryoet-alignment

Convert between different alignment formats used in cryo-ET.

Supported formats:

• IMOD
• AreTomo3
• Warp (global tilt-series alignment)
• RELION 5 (global tilt-series alignment)
• cryoet-data-portal

Installation

cryoet-alignment can be installed using pip:

pip install cryoet-alignment

Usage

Reading and writing alignment files

cryoet-alignment provides a simple API to read and write alignment files from different software packages.

IMOD

When processing tomography data using IMOD/etomo, files containing relevant alignment information are usually stored following the naming convention basename.xf (in-plane parameters), basename.tlt (tilt angles), basename.xtilt (x-rotation), basename.mrc (unaligned tilt series) and basename_full_rec.mrc (tomogram).

This layout is assumed when reading and writing IMOD alignment files as shown below. Any present tilt.com and newst.com file in the same directory may also be read.

from cryoet_alignment import read
from cryoet_alignment import write

# Read IMOD alignment files using etomo basename
imod_alignment = read("/path/to/imod_dir/basename")

# Write IMOD file
write(imod_alignment, "/path/to/imod_dir/basename")

AreTomo3

When processing tomography data using AreTomo3, alignment information is stored in a single .aln file. This file can be read and written as shown below.

from cryoet_alignment import read
from cryoet_alignment import write

# Read AreTomo3 alignment files
aretomo3_alignment = read("/path/to/alignment_file.aln")

# Write AreTomo3 file
write(aretomo3_alignment, "/path/to/alignment_file.aln")

AreTomo3's per-tilt CTF estimates (<name>_CTF.txt, one row per raw tilt including dark frames) can also be read and written. Because the .txt extension is generic, the reader is never inferred — pass reader="aretomo3_ctf" explicitly.

from cryoet_alignment import read
from cryoet_alignment import write

# Read AreTomo3 CTF estimates
aretomo3_ctf = read("/path/to/name_CTF.txt", reader="aretomo3_ctf")

# Write AreTomo3 CTF file
write(aretomo3_ctf, "/path/to/name_CTF.txt")

Warp

Warp stores tilt-series metadata (including the global alignment: tilt angles, per-tilt tilt-axis rotation, and 2D shifts in Å) in an XML file per tilt series. Only the global alignment is modeled; Warp's local warp grids have no analog in the other formats and are ignored. Because the .xml extension is generic, the reader is never inferred — pass reader="warp" explicitly. The tilt-image pixel size (Å/px) is needed to convert the Å shifts to pixels; if not given, it is read from the XML's CTF PixelSize parameter.

from cryoet_alignment import read
from cryoet_alignment import write

# Read a Warp tilt-series XML
warp_alignment = read("/path/to/tilt_series.xml", reader="warp", pixel_size_a=1.7005)

# Write a Warp tilt-series XML
write(warp_alignment, "/path/to/tilt_series.xml")

RELION 5

RELION 5 stores tilt-series metadata in a tomograms.star plus per-tomogram star files. Both on-disk layouts are read: the RELION-5 layout (rlnTomoTiltSeriesStarFile references) and the relion-4/WarpTools layout (per-tomogram blocks embedded in one file), including relion-4 projection matrices (rlnTomoProjX/Y/Z/W), which are decomposed into Euler angles and shifts exactly like RELION does it. Only the global alignment is modeled. Writing emits the RELION-5 two-file layout (tomograms.star + tilt_series/<name>.star). Pass tomo_name when the file lists several tomograms, and image_size_px when the per-tilt table carries matrices instead of Euler columns.

from cryoet_alignment import read
from cryoet_alignment import write

# Read a RELION tomograms.star (one tomogram)
relion_alignment = read("/path/to/tomograms.star", tomo_name="TS_01")

# Write the RELION-5 two-file layout
write(relion_alignment, "/path/to/out/tomograms.star")

cryoet-data-portal

Alignment information from the cryoet-data-portal is stored in a JSON file with a schema described here. This file can be read and written as shown below.

from cryoet_alignment import read
from cryoet_alignment import write

# Read cryoet-data-portal alignment files
cryoet_data_portal_alignment = read("/path/to/alignment_file.json")

# Write cryoet-data-portal file
write(cryoet_data_portal_alignment, "/path/to/alignment_file.json")

Convert between different alignment formats

cryoet-alignment provides the ability to convert between different alignment formats. For any conversion, the alignment object must be read first using the appropriate read function, and then converted to the cryoet-data-portal format before converting and writing to the desired format.

IMOD to AreTomo3

from cryoet_alignment import read, write
from cryoet_alignment.io.cryoet_data_portal import Alignment

# Read IMOD alignment files using etomo basename
imod_alignment = read("/path/to/imod_dir/basename")

# Convert IMOD to AreTomo3
cdp_alignment = Alignment.from_imod(imod_alignment)

# Write AreTomo3 file
tilt_series_dim = (4096, 4096, 41)
write(cdp_alignment.to_aretomo(ts_size=tilt_series_dim), "/path/to/alignment_file.aln")

AreTomo3 to Warp

from cryoet_alignment import read, write
from cryoet_alignment.io.cryoet_data_portal import Alignment

# Read the AreTomo3 alignment and convert to the canonical format
cdp_alignment = Alignment.from_aretomo3(read("/path/to/alignment_file.aln"), vol_size=(4096, 4096, 2000))

# Convert to Warp and write the tilt-series XML
warp_alignment = cdp_alignment.to_warp(pixel_size_a=1.54, image_size_px=(4096, 4096))
write(warp_alignment, "/path/to/tilt_series.xml")

AreTomo3 to RELION

from cryoet_alignment import read, write
from cryoet_alignment.io.cryoet_data_portal import Alignment

# Read the AreTomo3 alignment and convert to the canonical format
cdp_alignment = Alignment.from_aretomo3(read("/path/to/alignment_file.aln"), vol_size=(4096, 4096, 2000))

# Convert to RELION and write tomograms.star + tilt_series/TS_01.star
relion_alignment = cdp_alignment.to_relion("TS_01", pixel_size_a=1.54)
write(relion_alignment, "/path/to/out/tomograms.star")

cryoet-data-portal to IMOD

It is also possible to convert directly from the cryoet-data-portal client to IMOD/AreTomo format. This is demonstrated below. This requires additional dependencies to be installed using the following command:

pip install cryoet-alignment[cdp]

To convert from the cryoet-data-portal to IMOD, the below code can be used. Briefly, given a tomogram ID, the snippet fetches the alignment information from the cryoet-data-portal, reads the tilt series metadata, and converts the alignment to IMOD format. The resulting alignment files are written to the specified directory with the portal's run name as the base name.

import cryoet_data_portal as cdp
import zarr
from cryoet_alignment.io.cryoet_data_portal import Alignment
from cryoet_alignment import write

# Target tomogram ID
# This is an example from dataset 10004 (https://cryoetdataportal.czscience.com/runs/333)
TOMO_ID = 771

# Get the tomogram from the cryoet-data-portal
client = cdp.Client()
tomogram = cdp.Tomogram.get_by_id(client, TOMO_ID)

# Read cryoet-data-portal alignment from S3
cdp_ali = Alignment.from_s3(tomogram.alignment.s3_alignment_metadata)

# Get the tilt series metadata
#tilt_series = tomogram.alignment.tiltseries < currently unavailable due to a bug in the data portal client
tilt_series = tomogram.run.tiltseries[0]
pixel_size = tilt_series.pixel_spacing
dim_z, dim_y, dim_x = zarr.open(tilt_series.s3_omezarr_dir)['0'].shape

# Convert to IMOD format
imod_ali = cdp_ali.to_imod(ts_size=(dim_x, dim_y, dim_z), ts_spacing=pixel_size)
write(imod_ali, f"/tmp/test/{tomogram.run.name}")