Ribosome

Example 1: yeast ribosomes

In this example we used Warp, AreTomo3, easymode, Relion5, and M to reconstruct, denoise, segment, pick, and average the S. cerevisiae ribosome.

Dataset and computational resources

For this test we used 500 tilt series of plasma-FIB milled S. cerevisiae cells which Sebastian Tacke and colleagues at the MPI Dortmund shared with us. They are not yet available online - but you should be able to follow along with this example using other data sources as well, since easymode networks are general.
We used 4 NVIDIA RTX 4090 GPUs for most processing steps.

At the onset the data in this example consisted of just tilt series and mdocs (we did not use the gain references).

project_root/
├── frames/        # approximately 20.000 .eer files
│   ├── 20240731_l10t01_001_6.0_20240731_170707.eer
│   ├── 20240731_l10t01_002_8.5_20240731_170731.eer
│   └── ... 
└── mdocs/         # 500 .mdoc files
    ├── 20240731_l10t01.mdoc
    ├── 20240731_l10t02.mdoc
    └── ...

Step 1: tomogram reconstruction

easymode reconstruct --frames frames --mdocs mdocs --apix 1.56 --dose 2.05 --no_halfmaps

We now have 500 reconstructed tomograms at 10.00 Å/px in warp_tiltseries/reconstruction/.

Step 2: tomogram denoising

easymode denoise --data warp_tiltseries/reconstruction --output warp_tiltseries/reconstruction/denoised --mode direct --method n2n --gpu 0,1,2,3

This produced 500 denoised tomograms in warp_tiltseries/reconstruction/denoised/.

Step 3: ribosome segmentation

easymode segment ribosome --data warp_tiltseries/reconstruction/denoised --output segmented --tta 1 --gpu 0,1,2,3

We now have 500 ribosome segmentation volumes in segmented/*__ribosome.mrc.

Step 4: ribosome picking

easymode pick ribosome --data segmented --output coordinates/ribosome --binning 3 --size 2000000 --spacing 250

This created 500 .star files in coordinates/ribosome/, one per tomogram, containing a total of 109081 ribosome coordinates.

Step 5: exporting particles with WarpTools

conda activate warp
WarpTools ts_export_particles --input_directory coordinates/ribosome --input_pattern "*.star" --coords_angpix 10.0 --output_star relion/ribosome/particles.star --output_angpix 5.0 --box 96 --diameter 250 --2d --relative_output_paths

From here on, we followed the WarpTools tutorial for averaging apoferritin.

Final result

The final average focused on the large ribosomal subunit, after multiple rounds of refinement in M, reached an overall resolution of 3.4 Å and up to 2.6 Å in the best-resolved regions. This average was weighted per tilt and per tilt series (M EstimateWeights --resolve_items --resolve_frames), but did not require any density-based 3D classification.