daqplugin

name	daqplugin
description	Use DAQplugin for residue-wise DAQ quality scoring of protein atomic models in cryo-EM maps through ChimeraX. Trigger when users ask for DAQ scores, DAQ coloring, map-model compatibility, residue-wise local quality, amino-acid assignment quality, DAQ .npy files, DAQ B-factor export, DAQ sequence-shift arrows, or live DAQ monitoring during ISOLDE/manual model movement.

DAQplugin

Drive DAQplugin, a ChimeraX bundle for residue-wise local quality scoring of protein models in cryo-EM maps. DAQplugin can compute DAQ probability grids, color structures by DAQ metrics, monitor live coordinate changes, draw sequence-shift suggestion arrows, and export scores through B-factors.

Use this skill together with chimerax for batch invocation. Use isolde as well when the user wants live DAQ feedback during interactive ISOLDE refinement.

Choose The Workflow

1. Compute new DAQ scores from a map: run daqscore compute_grid when the user has a cryo-EM map and wants a reusable .npy score grid.
2. Score directly at atom/model positions: run daqscore compute_pdb when the user wants original-style DAQ scoring or a scored model quickly.
3. Color or export from existing scores: run daqcolor apply followed by save when the user already has a DAQ .npy.
4. Inspect sequence-register problems: run daq arrowwin when the user asks for sequence-shift suggestions or DAQ arrows.
5. Live monitoring: use the GUI/interactive ChimeraX plus ISOLDE. Do not promise live monitoring from a --nogui batch job.

Load references/commands.md when you need exact command syntax, metrics, backend names, or examples.

Installation Checks

DAQplugin is a ChimeraX bundle. Prefer the ChimeraX Toolshed install if available:

toolshed install /path/to/chimerax_daqplugin-X.Y.Z-py3-none-any.whl

For a development checkout:

devel clean /path/to/DAQplugin/daqcolor
devel install /path/to/DAQplugin/daqcolor
help daqcolor
help daqscore

The upstream repository uses submodules, so clone source checkouts with:

git clone --recurse-submodules https://github.com/kiharalab/DAQplugin.git

If command registration is uncertain, run a tiny ChimeraX job with help daqcolor or help daqscore and treat missing help as a failed install.

Batch Pattern

Use the chimerax skill wrapper for one-shot jobs. Start with close all, open map/model with explicit formats, run DAQplugin commands, save to new files, then inspect the wrapper result JSON and output files.

{
  "resultFile": "/tmp/daq_job/result.json",
  "commands": [
    "close all",
    "open /abs/path/map.mrc format mrc",
    "open /abs/path/model.cif format mmcif",
    "daqscore compute_grid #1 0.007 structure #2 metric aa_score output /abs/path/daq_scores.npy stride 2 half_window 9 backend auto",
    "save /abs/path/model_daq_colored.cif #2"
  ]
}

For a precomputed .npy:

daqcolor apply /abs/path/daq_scores.npy #2 metric aa_score half_window 9
save /abs/path/model_daq_colored.cif #2

Prefer .cif output. DAQ scores used for coloring are written into the model B-factor field when the model is saved.

Practical Defaults

• Use aa_score first for amino-acid assignment quality.
• Use atom_score for CA/atom-position support.
• Use ss_score for secondary-structure agreement when available.
• Use half_window 9 for normal residue-wise smoothing unless the user asks for a sharper local view.
• Use stride 2 for grid scoring unless the user prioritizes maximum detail over runtime.
• Use backend auto first. Force cpu for reproducibility or when GPU backends fail.
• Set batch_size lower if the run hits GPU/host memory limits.
• Save both the .npy and a colored/scored .cif into the job output folder.

Backend And Runtime Notes

DAQplugin auto-selects inference backends by platform:

Platform	Typical chain
Linux NVIDIA	TensorRT -> CUDA -> CPU
Windows	DirectML -> CPU
macOS Apple Silicon	MLX-Metal -> MLX-CPU -> ORT-CPU
macOS Intel	ORT-CPU

The active backend is printed in the ChimeraX log. If a forced GPU backend silently falls back or fails, rerun with backend cpu to separate DAQ/plugin correctness from GPU setup.

ISOLDE / Live Monitoring

For live feedback during model movement:

1. Compute or load an existing DAQ .npy.
2. In interactive ChimeraX, start DAQplugin from Tools > Validation > DAQplugin, or run daqcolor monitor.
3. Move/refine in ISOLDE.
4. Stop monitoring before closing or saving final outputs:

daqcolor monitor #2 npy_path /abs/path/daq_scores.npy metric aa_score interval 0.5
daqcolor monitor #2 on false

Batch --nogui jobs are appropriate for compute/apply/save, not for interactive monitoring.

Sequence-Shift Arrows

Use DAQ arrows when the task is to find possible residue-register or sequence-shift errors:

daq arrowwin #2 /abs/path/daq_scores.npy nwin 5 kshift 5 min_improvement 0.5
daq arrowwin #2 /abs/path/daq_scores.npy chain A apply_isolde_restraints true spring_constant 1500
daq clearrestraints #2

Only use apply_isolde_restraints true when the user explicitly wants ISOLDE restraints from DAQ suggestions.

Interpretation

• Positive DAQ values mean local density supports the modeled amino-acid type better than the average distribution.
• Negative values suggest possible amino-acid misassignment or local model inconsistency.
• Near-zero values are ambiguous and can reflect low local resolution or weak density.
• Treat DAQ as evidence for inspection and rebuilding, not as an automatic model-editing instruction.

Failure Modes

• Wrong map contour: grid scoring samples points above the contour. Use the same contour level used for map inspection unless the user gives a different threshold.
• Map/model frame mismatch: DAQ scores are meaningless if the model and map are not in the same coordinate frame.
• Missing .npy output: treat the run as failed even if ChimeraX exits cleanly.
• Plugin not registered: daqscore/daqcolor commands will be unknown. Check Toolshed/dev install.
• Memory pressure: reduce batch_size, increase stride, cap max_points, or force backend cpu.
• Monitoring in batch: live coordinate tracking expects interactive ChimeraX frame updates; use GUI/ISOLDE, not --nogui.

Citation

When reporting DAQ results, cite Terashi et al., Nature Methods 2022, "Residue-wise local quality estimation for protein models from cryo-EM maps", and note that DAQplugin came from https://github.com/kiharalab/DAQplugin.