CryoSPARC Guide
Search
K

Job: Particle Subtraction

At a Glance

Subtract projections of a masked sub-volume from particle images.
  • Use input particles with a gold-standard half set
  • Ensure that the region to be subtracted is well-aligned
  • The mask must have a soft edge

Description

A particle subtraction job is used to remove unwanted signal from particle images. The input volume is projected in the fitted pose for each input particle. This projection is then subtracted from the particle image, yielding an image of the particle as if it did not have the masked out region.
Because the input volume is subtracted from the particle images, the final quality of the signal subtraction is highly dependent on the quality of the input volume. Often, performing a Local Refinement of the sub-volume that is to be subtracted prior to performing Particle Subtraction improves the quality of this region, which therefore improves the quality of the final particle stack.
In a particle subtraction job, a sub-volume is projected in each particle image's pose. This projection is then subtracted from the image, resulting in an image which ideally contains only signal from the region outside the mask.

Inputs

Particles

Particles must be from a previous refinement job with half-set splits (i.e., homogeneous or non-uniform refinement). Results may improve if per-particle scale factors have been optimized.
Why does Particle Subtraction require half-sets?
In particle subtraction, a volume is subtracted from all particle images. If this volume was reconstructed from all particle images, the resulting subtracted particle stack would contain information from all particles, breaking the gold-standard assumption of half-set independence. Thus, each particle must only have the half-map from its respective half-set subtracted. The particle subtraction job will internally ensure that this is satisfied.

Volume

Volume must include both half maps. The volume must be the same size as the particle images.

Mask

The mask should cover the region of the volume that is to be subtracted from the images. Like other masks which “cut through” map density, the mask must have a soft edge. However, if your mask is too large it may incorporate noise from regions of the map with no appreciable signal. Thus, the size of the mask must be carefully tuned.
We recommend a minimum soft padding width of
5×resolutionapix5 \times{} \frac{\mathrm{resolution}}{\mathrm{apix}}
where resolution is the volume’s GSFSC resolution in Å and apix is the volume’s pixel size in Å, but you may need to test several masks to find the optimal result. More advice on mask making is available in Mask Creation.

Common Problems

Windowing and scaling

It is essential that the windowing parameters in a Particle Subtraction job match those of the input refinement. The volume is multiplicatively scaled to match the contrast of each particle image, and using different windowing parameters may result in an incorrect scaling factor.

Common Next Steps

We recommend performing a Homogeneous Reconstruction Only using the subtracted particle sets as a sanity check that the correct parts of the image have been subtracted.
If the sub-volume that was subtracted was aligned prior to performing the subtraction, it can be helpful to use the low-level interface to replace the poses of the subtracted particles with those from an earlier refinement in which the remaining density was properly aligned.
Particle subtracted images are often most useful in Local Refinement.