Is it possible to extract particles of each frame of 3D Variability display component?

Hi,
Is there a way to extract particles of each frame from the output component of 3D variability Display (Simple or intermediate)? Want to extract and further refine those particles separately.

Best

You can use 3DVA display with cluster mode then refine the output particles. I believe you can restrict the clustering to one component by modifying the input with the lower-level results output (ie plot component 0 v component 0, rather than 0 v 1 v 2).

3 Likes

Hi @rmattoo,

Thanks for expressing interest in this functionality. The only way right now that particles can be separated into distinct subsets via 3DVA Display is by using the cluster mode. The output of a 3DVA Display job in cluster mode will be subset stacks and reconstructed maps for each cluster. Currently we haven’t implemented the same per-frame subsetting functionality in the intermediate mode, but this is on the roadmap.

Best,
Michael

1 Like

Hi @mmclean,
Thanks for the response. One more question, If I do 3DVA Display job in ‘cluster’ mode with let’s say three clusters (000, 001 and 002). I extract the three particle stack files and do 3DVA Display job in ‘Intermediate’ mode with three components (000, 001 and 002). Would the cluster 000 correspond to component 000, cluster 001 correspond to component 001 and so on?

Hi @rmattoo,

If I understand correctly, no. The cluster, simple, and intermediate modes are three independent ways of looking at the same information: the latent variables and their associated eigenvectors (the 3D components themselves).

The cluster mode attempts to find the best-fitting set of K clusters that fit the distribution of latent variables in the dataset (K is totally independent of the number of components used in the previous 3DVA job). The intermediate mode instead takes each volume component, assigns particles into L bins based on their latent variables along that component, and then performs L reconstructions per component, where only the binned particles contribute to each reconstruction. If you selected cluster mode, the number of frames/clusters parameter is K; if you selected intermediate mode, it instead corresponds to L.

The reason that one parameter is used for these independent quantities is just for implementation convenience – there’s no relation or correspondence between the fitted clusters, and the components in the simple/intermediate modes. Part 2 of our 3DVA tutorial covers the cluster/intermediate modes in more detail. Let me know if this helps clarify :slight_smile:

Best,
Michael