bytedancebytedanceOfficialยท1 Agent Skills Included

PXMeter

Evaluate biomolecular structure predictions with standardized quality metrics

Evaluates protein, nucleic acid, and small molecule structure predictions against reference CIF files using LDDT, DockQ, RMSD, and PoseBusters checks. Eliminates manual chain matching and inconsistent scoring when comparing models like AlphaFold3, Boltz, and Protenix. Runs single-file evaluations or full dataset benchmarks and aggregates results into clear summary tables.
npx skills add bytedance/PXMeter --all -g -y

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install PXMeter?โ–ผ

Run `npx skills add bytedance/PXMeter --all -g -y` in your terminal to install this skill globally.

How to check if a predicted protein structure is accurate?โ–ผ

Run `pxm -r reference.cif -m model.cif -o output.json` to compare the prediction against the reference. PXMeter reports LDDT, DockQ, and RMSD scores so you can judge prediction quality at a glance.

How to compare AlphaFold3, Boltz, and Protenix results?โ–ผ

PXMeter includes built-in evaluators for these models. Run the benchmark workflow on your prediction folders, then aggregate everything into a single summary CSV with success rates for each model.

Why is ligand RMSD missing from my PXMeter results?โ–ผ

Ligand metrics are not calculated by default. Add the `-l` flag with the ligand chain IDs (e.g. `-l A,B`) to enable pocket-aligned RMSD and PoseBusters validity checks.

Does shuffled chain order affect PXMeter scores?โ–ผ

No. PXMeter automatically matches chains and atoms between the reference and model using sequence alignment and geometric symmetry resolution, so chain ordering never distorts the scores.

Related Repositories in Education & Research

View All in Education & Researchโ†’