13. High resolution 3D refinement

In this step, we’ll do a high resolution 3D refinement using the particles that we obtained from the last particle selection step. The high resolution refinement needs subtomograms and a reference obtained at bin 1 level. Below tables summarizes our input parameters for the Extract, Reconstruct particle, and the 3D refinement jobs.

13.1. Extraction of pseudosubtomograms (relion.pseudosubtomo)

Input params.	Bin 1
Input optimisation set:
Input tomogram set:	Tomograms/job006/tomograms.star
Input particle set:	Select/job017/particles.star
Box size (binned pix):	512
Cropped box size (binned pix):	192
Binning factor:	1
Outputs:
Pseudosubtomograms star file:	Extract/job018/optimisation_set.star

13.2. Reconstruct particle (relion.reconstructparticletomo)

Input params.	Bin 1
Input optimisation set:	Extract/job018/optimisation_set.star
Input tomogram set:
Input particle set:
Box size (binned pix):	512
Cropped box size (binned pix):	192
Binning factor:	1
Symmetry:	C6
Outputs:
3D reference density map:	Reconstruct/job019/merged.mrc
half1 density map:	Reconstruct/job019/half1.mrc
half2 density map:	Reconstruct/job019/half2.mrc

13.3. 3D refinement (auto-refine)

Input params.	Bin 1
Input optimisation set:	Extract/job018/optimisation_set.star
Reference map:	Reconstruct/job019/half1.mrc
Reference mask (optional):	align_mask.mrc (note below)
Ref. map is on absolute greyscale?	Yes
Resize reference if needed?	Yes
Initial low-pass filter (A):	5.5
Symmetry:	C6
Do CTF-correction?	Yes
Ignore CTFs until first peak?	No
Mask diameter (A):	230
Mask individual particles with zeros?	Yes
Use solvent-flattened FSCs?	Yes
Use Blush regularisation?	No
Initial angular sampling:	1.8
Initial offset range (pix):	5
Initial offset step (pix):	1
Local searches from auto-sampling:	1.8
Relax symmetry:
Use finer angular sampling faster?	No
Prior width on tilt angle (deg):	10
Use parallel disc I/O?	Yes
Number of pooled particles:	30
Skip padding?	No
Skip gridding?	Yes
Pre-read all particles into RAM?	No
Copy particles to scratch directory:
Combine iterations through disc?	No
Use GPU acceleration?	Yes
Number of MPI procs:	5
Number of threads:	8
Outputs:
Achieved Resolution (A)	5.18
Refined particles	Refine3D/job020/run_data.star
Optimisation set	Refine3D/job020/run_optimisation_set.star
Refined averaged density map:	Refine3D/job020/run_class001.mrc
half1 density map:	Refine3D/job020/run_half1_class001_unfil.mrc
half2 density map:	Refine3D/job020/run_half2_class001_unfil.mrc

Reference mask

In the high-resolution 3D refinement, we often use a mask to improve the final alignment. The alignement mask covers the region of interest, and in this case, the diameter of the mask is roughly about 230 A and it covers only the HIV capsid and excludes the matrix. We have used a disc-shaped mask generated from a custom Python script (explained elsewhere) but you may use other tools to generate your own masks. As the mask is used in the alignement, make it simple and less detailed. We include below screenshots for your reference.

Orthogonal views of alignment mask (align_mask.mrc)

Solvent mask

In the Postprocessing step below, we use another mask called fsc_mask.mrc to exclude the solvent area from the HIV hexemer. This covers only the central hexemer as you see in the screenshots below:

Orthogonal views of solvent mask (fsc_mask.mrc)

13.4. Removing further duplicate particles

During refinement, particles’s positions change. It is a good idea to furhter check for duplicated particles. We have removed duplicated particles within 50 A radius after the high resolution refinement.

13.5. Removal of duplicates (relion.select.removeduplicates)

Input params.	Values
OR select from particles.star:	Refine3D/job020/run_data.star
Minimum inter-particle distance (A)	50
Outputs:
Particle star file:	Select/job021/particles.star

13.6. Reconstruct particle (relion.reconstructparticletomo)

Input params.	Values
Input optimisation set:
Input tomogram set:	Tomograms/job006/tomograms.star
Input particle set:	Select/job021/particles.star
Box size (binned pix):	512
Cropped box size (binned pix):	192
Binning factor:	1
Symmetry:	C6
Outputs:
3D reference density map:	Reconstruct/job022/merged.mrc
half1 density map:	Reconstruct/job022/half1.mrc
half2 density map:	Reconstruct/job022/half2.mrc

13.7. Post processing (relion.postprocess)

Input params.	Values
One of the 2 unfiltered half-maps:	Reconstruct/job022/half1.mrc
Solvent mask:	fsc_mask.mrc
Outputs:
Postprocessed map:	Postprocess/job023/postprocess.mrc
Postprocessed masked-map:	Postprocess/job023/postprocess_masked.mrc

At the end of post-processing we obtained a map with 5.08 A resolution as confirmed by the gold-standard FSC curve below.

Left- FSC curve; Right- Density view of the postprocessed-masked map

We refer you to the Relion documentation for more details.

Relion 3D refinement documentation.