Perform 3D classification

After you upload the extracted particle stacks and files downloaded from Data Dryad to an Amazon EBS volume, boot up a cluster with the EBS volume mounted (read more about how to do that [here] (https://sites.google.com/site/emcloudprocessing/home/)). Essentially you need all files except the micrographs.

Once on the cluster, submit the 3D classification job to the cluster using the script run_3DClass.run:

#!/bin/bash

#$ -cwd
#$ -S /bin/bash
#$ -pe orte 256
#$ -q all.q
#$ -V
#$ -N run_3Dclass

mpirun relion_refine_mpi --o Class3D/run1_3dclass --i particles.star --particle_diameter 350 --angpix 1.77 --ref emd_2275_scaled_bin4_filt60A_sca_240.mrc --firstiter_cc --ctf --iter 15 --tau2_fudge 4  --K 4 --flatten_solvent --zero_mask --oversampling 1 --healpix_order 2 --offset_range 5 --offset_step 2 --sym C1 --norm --scale  --j 1 --memory_per_thread 4 --dont_combine_weights_via_disc

After 13 iterations, you'll see that the resolution and 3D volumes converge. To see an example of this, you can find the 13th iteration of my classification:

Class3D/run1_3dclass_it013_*

At this point, you will continue the refinement using local angular searches (within 10 degrees) and a smaller angular sampling (1.8 degrees). Submit job run_3DClass_local.run:

#!/bin/bash

#$ -cwd
#$ -S /bin/bash
#$ -pe orte 256
#$ -q all.q
#$ -V
#$ -N run_3Dclass_local

mpirun relion_refine_mpi --o Class3D/run1_ct13 --continue Class3D/run1_3dclass_it013_optimiser.star --iter 25 --tau2_fudge 4 --oversampling 1 --healpix_order 4 --sigma_ang 3.33333 --offset_range 5 --offset_step 2 --j 1 --dont_combine_weights_via_disc

Let this run until it finishes at 25 iterations.

For my data, the first and third classes looked to be within the same conformation. Classification data:

Class3D/run1_ct13_it025_*

So I merged these two classes into a single .star file:

Class3D/run1_ct13_it025_data_class1_and3.star