Skip to content

Processing Tilt Series Stacks

WarpTools expects multi-frame micrographs (movies) as input when processing tilt series data. Sometimes only tilt series stacks are available. This page will show you how to prepare tilt series stacks for processing in WarpTools.

We will use a SARS-CoV-2 tilt series from EMPIAR-10453 to demonstrate. The script may need to be modified for other datasets.

Splitting Tilt Series Stacks

We will use a simple script to split our tilt series into individual tilt images.

Preparation

This Python scripts depends on the Python packages mrcfile, tifffile and mdocfile. These can be installed from PyPI in your warp conda environment.

pip install mrcfile tifffile mdocfile

Running the script

Below is the starting directory structure and the contents of the Python script.

directory structure (single tilt series)
.
├── mdoc
│   └── TS_005.mrc.mdoc
├── tiltseries
│   └── TS_005.mrc
└── warp_ts_split_10453.py
warp_ts_split_10453.py
from pathlib import Path
import mrcfile
import mdocfile

# modify these variables for your own processing
OUTPUT_DIRECTORY = 'movies'
TILT_SERIES_DIRECTORY = 'tiltseries'
MDOC_DIRECTORY = 'mdoc'


# define a function for processing a single tilt series
def process_tilt_series(mdoc_file: Path, output_directory: Path):
    # load mdoc into pandas dataframe
    df = mdocfile.read(mdoc_file)

    # load tilt series into numpy array
    tilt_series_file = Path(TILT_SERIES_DIRECTORY) / mdoc_file.stem
    tilt_series = mrcfile.read(tilt_series_file)

    # check for same number of tilts in mdoc and image file
    if len(df) != len(tilt_series):
        e = f'more tilts in mdoc than tilt series for {mdoc_file}'
        raise RuntimeError(e)

    # split tilt series into individual tilts
    for movie_path, image in zip(df['SubFramePath'], tilt_series):
        output_movie_file = output_directory / Path(movie_path.name)
        if output_movie_file.suffix.lower() == '.mrc':
            mrcfile.write(output_movie_file, image)
        elif output_movie_file.suffix in ('.tif', '.tiff'):
            tifffile.imwrite(output_movie_file, image)

# main program which loops over all tilt series...
if __name__ == '__main__':
    # create output directory
    Path(OUTPUT_DIRECTORY).mkdir(exist_ok=True, parents=True)

    # get list of mdoc files to process
    mdocs = list(Path(MDOC_DIRECTORY).glob('*.mdoc'))

    # process mdoc files one by one
    for mdoc in mdocs:
        process_tilt_series(mdoc_file=mdoc, output_directory=Path(OUTPUT_DIRECTORY))

python warp_ts_split_10453.py

Output

output directory structure
.
├── mdoc
│   └── TS_005.mrc.mdoc
├── movies
│   ├── TS_005_001_-0.0.tif
│   ├── TS_005_002_3.0.tif
│   ├── TS_005_003_6.0.tif
│   ├── TS_005_004_-3.0.tif
│   ├── TS_005_005_-6.0.tif
│   ├── TS_005_006_9.0.tif
│   ├── TS_005_007_12.0.tif
│   ├── TS_005_008_-9.0.tif
│   ├── TS_005_009_-12.0.tif
│   ├── TS_005_010_15.0.tif
│   ├── TS_005_011_18.0.tif
│   ├── TS_005_012_-15.0.tif
│   ├── TS_005_013_-18.0.tif
│   ├── TS_005_014_21.0.tif
│   ├── TS_005_015_24.0.tif
│   ├── TS_005_016_-21.0.tif
│   ├── TS_005_017_-24.0.tif
│   ├── TS_005_018_27.0.tif
│   ├── TS_005_019_30.0.tif
│   ├── TS_005_020_-27.0.tif
│   ├── TS_005_021_-30.0.tif
│   ├── TS_005_022_33.0.tif
│   ├── TS_005_023_36.0.tif
│   ├── TS_005_024_-33.0.tif
│   ├── TS_005_025_-36.0.tif
│   ├── TS_005_026_39.0.tif
│   ├── TS_005_027_42.0.tif
│   ├── TS_005_028_-39.0.tif
│   ├── TS_005_029_-42.0.tif
│   ├── TS_005_030_45.0.tif
│   ├── TS_005_031_48.0.tif
│   ├── TS_005_032_-45.0.tif
│   ├── TS_005_033_-48.0.tif
│   ├── TS_005_034_51.0.tif
│   ├── TS_005_035_54.0.tif
│   ├── TS_005_036_-51.0.tif
│   ├── TS_005_037_-54.0.tif
│   ├── TS_005_038_57.0.tif
│   ├── TS_005_039_60.0.tif
│   ├── TS_005_040_-57.0.tif
│   └── TS_005_041_-60.0.tif
├── tiltseries
│   └── TS_005.mrc
└── warp_ts_split_10453.py

what if I don't have mdoc files?

You'll have to make them yourself! The mdocfile Python package might be useful.

Subsequent Processing

You should now be able to follow the quick start guide for tilt series processing.

Differences

  • use fs_ctf not fs_motion_and_ctf - you don't need to correct motion!
  • copy and paste or symlink motion corrected averages into an averages directory inside your frame series processing directory
  • you won't be able to refine motion correction results in M