This is for the new subtomogram-subtilt refinement pipeline that is available in Github and continuous build after EMAN2.91 (03/09/2021). Both the pipeline and the tutorial are still under construction. Backward compatibility of the results is not guaranteed...

Tomogram reconstruction

Particle selection

Subtomogram-subtilt refinement

To use, simply run

With default options, the tutorial dataset should be able to get to ~7.5Å resolution. While the reported number is lower, the features in the resulting map are about as good as EMD-11654.

Visualize particle motion in individual tilts

After a refinement, run --path spt_xx --loadali2d <last aliptcls2d_xx.lst file> --loadali3d <last aliptcls3d_xx.lst file> 

It will take a while to load all metadata, and plot the trajectory of each particle on each tilt image in each tomogram.

In the top panel, the blue curve represents the average score of all 2D particles in that tilt, and the red curve represents the average distance of the subtilt motion with respect to the alignment of the 3D particle. The quiver plot below shows the trajectory of each individual particle, colored by its alignment score.

Focused refinement and continuous motion

After a refinement run, make a customized mask for the domain of interest using the filter tool, then generated the masked referenced. Make sure to mask both the even and odd maps that will be used as references and name them properly. spt_xx/threed_yy_even.hdf spt_xx/threed_yy_masked_even.hdf --multfile mask_smallunit.hdf spt_xx/threed_yy_odd.hdf spt_xx/threed_yy_masked_odd.hdf --multfile mask_smallunit.hdf

Do another short run with the masked reference with the --localrefine option. Load the last aliptcls2d and aliptcls3d file from the last run use the corresponding options. In this case, using --iters=p,t will be enough. --ptcls sets/ptcls.lst --ref spt_xx/threed_yy_masked.hdf --iters p,t --startres 7 --goldcontinue --loadali2d spt_xx/aliptcls2d_yy.lst --loadali3d spt_xx/aliptcls3d_yy.lst --mask mask_smallunit.hdf --localrefine

Then, run --path spt_xx --ali3dold spt_xx/aliptcls3d_00.lst --ali3dnew spt_xx/aliptcls3d_01.lst --ali2d spt_xx/aliptcls2d_00.lst

Note that since we want to look at the motion of the masked domain within the coordinates of the original refinement, here use the old aliptcls2d.lst file. If the goal is to visualize the motion of other parts of the protein with the focused part fixed, use the newer aliptcls2d file. The program will output two 3D map stacks corresponding to the trajectories of the first two eigenvectors. They can be filtered/masked manually with before visualization.

Multi-reference refinement

To run a classical multi-reference refinement (simultaneous alignment and classification), use ref1.hdf ref2.hdf --ptcls sets/ptcls.lst --niter 5

You can also provide only one reference and ask for N classes using the --nref option. The program will phase randomize the given map to create N initial references. ref1.hdf --ptcls sets/ptcls.lst --niter 5 --nref 3

Note that the multi-reference refinement will NOT follow the "gold-standard" protocol, so it is highly recommended to specify a resolution cutoff using the --maxres option (default is 20A). The program also will not apply masks to the output maps, but it will mask the reference at the beginning of each iteration if a mask file is provided through --maskalign.

From an existing single model refinement, to run an iterative classification without alignment, run ref1.hdf ref2.hdf --ptcls spt_xx/aliptcls3d_xx.lst --niter 5 --skipali --loadali3d

Make sure to replace use the aliptcls3d_xx.lst from the single model refinement as particle input, and enable the --loadali3d option. The --skipali option here can be omitted, and the program will do a local search around the given solution in this case.

Also when --loadali3d is specified, you can also simply ask for N classes without providing any reference maps. i.e. --ptcls spt_xx/aliptcls3d_xx.lst --niter 5 --skipali --loadali3d --nref 3

In this case, the program will classify particles to random classes at the beginning of the refinement, and reconstruct maps from the classes to use as initial references.

Symmetry breaking can also be performed using this program under the --loadali3d mode. Note that here we require a symmetry string to be specified using the --breaksym option, so you can break a c6 symmetry to c3 using --breaksym c2 --sym c3. Similar to the classification process, the program will randomly assign particles to different asymmetrical units at the beginning of the refinement.

Initial model generation

After 08/2021, there is a program for new initial model generation in the subtomogram averaging protocol. It should be more robust than the previous version, and supports simultaneous particle classification during initial model generation. For a simple run to build an initial model at 50Å resolution, set/ptcls.lst --res 50

To generate two initial models from a set of particles, run set/ptcls.lst --res 50 --ncls 2

Here is an example of initial model generation from a dataset of microtubule-shaped particles from tomograms of flagella. The program can classify the particles generate the initial model of microtubule doublet and the central pair simultaneously.

The final output can be found in sptsgd_xx/output_clsx.hdf, and the intermediate maps can be found in sptsgd_xx/output_all_clsx.hdf. Note that not all particles are used in the initial model generation, so there won't be the class assignment of the particles. To get that, run the multi-model refinement with the output maps as references.

EMAN2/e2tomo_new (last edited 2021-09-24 19:31:54 by MuyuanChen)