|
|
The script [sim_spinifel.py](https://github.com/apeck12/cmtip/blob/main/examples/sim_spinifel.py) in cmtip/examples can be used to generate new datasets. For the chaperonin (PDB ID: 3iyf) simulated on a small monolithic detector of 128 x 128 pixels, we recommend running the script with the following parameters:
|
|
|
|
|
|
`python sim_spinifel.py -b amo86615.beam -p 3iyf.pdb -d 128 0.08 0.2 -n 500000 -s 10 -o 3iyf_clean.h5`
|
|
|
|
|
|
This will generate a 500k image dataset of noise-free diffraction intensities. In this case, the resolution at the edge of the detector, which dictates the high-resolution limit of the reconstruction, is 13.8 Å. To add Poisson noise, the above command should be modified as follows:
|
|
|
|
|
|
`python sim_spinifel.py -b amo86615.beam -p 3iyf.pdb -d 128 0.08 0.2 -n 500000 -s 10 -q -o 3iyf_poisson.h5`
|
|
|
|
|
|
where the `-q` flag results in the diffraction intensities being quantized to simulate photon counts on the detector. The overall fluence is increased from the the default given in amo86615.beam by the `-s` flag, resulting in a total fluence of 1e13 photons per diffraction shot. Additional types of noise, including beam and fluence jitter, can be simulated as follows:
|
|
|
|
|
|
`python sim_spinifel.py -b amo86615.beam -p 3iyf.pdb -d 128 0.08 0.2 -n 500000 -s 10 -q -bj 1 -fj 0.2 -o 3iyf_noisy.h5`
|
|
|
|
|
|
A custom sloped background, which results in systematic rather than uncorrelated errors between shots, can also be simulated by supplying the path to a numpy array of the desired background and using the `-sb` flag.
|
|
|
|
|
|
The sim_spinifel.py file requires a single GPU, and the beam file can be found at: [skopi/examples/input/beam](https://github.com/chuckie82/skopi/blob/main/examples/input/beam/amo86615.beam). A representative image from each dataset (with the particle in a random orientation for each shot) is shown below. Although the Poisson and noisy images are visually indistinguishable, the latter is affected by beam and fluence jitter, with the beam's center displaced by 2.2 pixels and a fluence of 1.2e13 rather than 1e13 photons for this shot. |