opus-et-agent

Cryo-ET data processing pipeline using WARP, AreTomo2, PyTOM, and OPUS-ET. Use this skill whenever the user is processing cryo-electron tomography (cryo-ET) data, including: importing frame series or tilt series from MDOC files, CTF estimation, AreTomo2 alignment, template matching with PyTOM, exporting subtomograms, or training OPUS-ET heterogeneity models. Also use when the user mentions tomostar files, WarpTools commands, tilt stacks (.st, .rawtlt), defocus handedness, ts_export_particles, or SLURM job submission for GPU tomography processing.

Supervised-autonomy orchestrator for the cryo-ET pipeline. Drives opus-et-warp (reconstruction) and opus-et-analysis (interpretation) end-to-end over SLURM, tracking progress in .opus_run_state.json, pausing at scientific checkpoints, and generating in-cell visualizations via opus-et-visualize. Use when the user wants to run, monitor, resume, or checkpoint a full cryo-ET run rather than a single phase.

Cryo-ET data processing and analysis workflows using opus-et training results. Handles PCA/kmeans clustering, volume generation from latent codes, pose parsing, and star file manipulation. Use when processing training results from a specific epoch, generating volumes for cluster centers or principal components, parsing poses, or combining star files from cryo-ET reconstructions.

Generate in-cell molecular visualizations for cryo-ET results. Two modes — (1) place a refined/averaged map at every particle pose inside its original tomogram in ChimeraX/ArtiaX, colored by OPUS-ET conformational state (the finale look); (2) REVEAL the raw density instead of replacing it — mark picks on the raw tomogram (per-particle zoomed gallery via particle_gallery.py, or slab overlays via tm_picks_overlay.py) with ring/transparent/solid markers, and scan the slice through Z. Use when the user wants molecules in cellular context, a hero in-cell render, or to show/validate that picks land on real raw density.