susan.project.SubtomoAvgSched

Warning

Experimental / in development. The scheduler protocol, built-in schedulers in susan.project.Schedulers, and the factory presets (make_*_refinement) may change between releases. Use SubtomoAvg directly for stable manual loops.

class susan.project.SubtomoAvgSched.SubtomoAvgSched(prj_name, box_size=None)[source]

Bases: SubtomoAvg

Subtomogram averaging with optional per-iteration schedulers.

Warning

Experimental / in development. The scheduler protocol and the factory presets may change between releases. Use SubtomoAvg directly for stable manual loops.

Extends SubtomoAvg with three scheduler slots and a run() loop. Leaving a scheduler as None keeps that setting unchanged each iteration.

Parameters:

  • prj_name (str) – Project directory.

  • box_size (int, optional) – Box size in pixels. Reads info.prjtxt when omitted.

bandpass_scheduler

Signature: new_bp = sched(ite, bp_est, current_bandpass). Result is assigned to both aligner.bandpass and ctf_refiner.bandpass. Default: None.

Type:

callable or None

angular_search_scheduler

Signature: (cs, css, is_, iss) = sched(ite, lp). Result is unpacked into aligner.set_angular_search(*result). Default: None.

Type:

callable or None

iteration_type_scheduler

Signature: type_str = sched(ite). Result is assigned to self.iteration_type. Default: None.

Type:

callable or None

Automated Loop

run(start, end, reset=False)[source]

Run iterations from start to end (inclusive).

Schedulers are applied before each iteration. bp_est passed to the bandpass scheduler is the resolution estimate (Fourier pixels) returned by the previous iteration (or aligner.bandpass.lowpass for the very first call).

Parameters:

  • start (int) – First iteration number.

  • end (int) – Last iteration number (inclusive).

  • reset (bool, optional) – If True, call reset() on all schedulers that support it before starting the loop. Default: False.

Returns:

Per-iteration resolution estimates in Fourier pixels, one entry per iteration.

Return type:

list of float or numpy.ndarray

Factory Presets

classmethod make_3d_refinement(prj_name, box_size=None, *, lp_init=30.0, lp_max=None, lp_step=2.5, cone_factor=4, inplane_factor=4, tomogram_file=None, initial_reference=None, initial_particles=None)[source]

3-D closed-loop refinement with adaptive angular search.

Bandpass chases the FSC estimate (clamped by lp_max); angular search spans scale with the current lowpass. First iteration uses no angular search (skip_first=True).

Parameters:

  • prj_name (str)

  • box_size (int, optional)

  • lp_init (float, optional) – Starting lowpass in Fourier pixels. Default: 30.

  • lp_max (float, optional) – Bandpass ceiling. Defaults to box_size // 2 - rolloff.

  • lp_step (float, optional) – Maximum lowpass increase per iteration. Default: 2.5.

  • cone_factor (float, optional) – Angular span multipliers. Default: 4.

  • inplane_factor (float, optional) – Angular span multipliers. Default: 4.

  • tomogram_file (str, optional) – Written to info.prjtxt only when provided (existing project files are left untouched otherwise).

  • initial_reference (str, optional) – Written to info.prjtxt only when provided (existing project files are left untouched otherwise).

  • initial_particles (str, optional) – Written to info.prjtxt only when provided (existing project files are left untouched otherwise).

Return type:

SubtomoAvgSched

classmethod make_2d_refinement(prj_name, box_size=None, *, lp_init=45.0, lp_max=None, lp_step=2.5, inplane_factor=4, tomogram_file=None, initial_reference=None, initial_particles=None)[source]

2-D in-plane refinement with adaptive inplane angular search.

Parameters:

  • prj_name (str)

  • box_size (int, optional)

  • lp_init (float, optional) – Default: 45.

  • lp_max (float, optional)

  • lp_step (float, optional) – Default: 2.5.

  • inplane_factor (float, optional) – Default: 4.

  • tomogram_file (str, optional)

  • initial_reference (str, optional)

  • initial_particles (str, optional)

Return type:

SubtomoAvgSched

classmethod make_mixed_ctf_2d(prj_name, box_size=None, *, lp_init=45.0, lp_max=None, lp_step=2.5, cone_factor=4, inplane_factor=4, odd='ctf', even='2D', tomogram_file=None, initial_reference=None, initial_particles=None)[source]

Interleaved CTF + 2-D refinement with shared bandpass tracking.

Parameters:

  • prj_name (str)

  • box_size (int, optional)

  • lp_init (float, optional) – Default: 45.

  • lp_max (float, optional)

  • lp_step (float, optional) – Default: 2.5.

  • cone_factor (float, optional) – Default: 4.

  • inplane_factor (float, optional) – Default: 4.

  • odd (str, optional) – Iteration types for odd and even iterations. Default: 'ctf' / '2D'.

  • even (str, optional) – Iteration types for odd and even iterations. Default: 'ctf' / '2D'.

  • tomogram_file (str, optional)

  • initial_reference (str, optional)

  • initial_particles (str, optional)

Return type:

SubtomoAvgSched