anonymous-motif-scaffolding/prosite_functional_motif_scaffolding_benchmark

PROSITE-derived Functional Motif Benchmark

This archive contains an anonymized dataset artifact for a systematically derived benchmark of structurally conserved functional motif-scaffolding cases from PROSITE-linked experimental protein structures.

The benchmark is intended for static motif-scaffolding evaluation with standard MotifBench-style pipelines. Cases are derived from PROSITE motif-pattern entries, mapped to experimentally resolved PDB structures, filtered for recurrent three-dimensional conservation across structural instances, and exported as motif-scaffolding tasks.

Contents

prosite_functional_motif_benchmark/
  README.md
  LICENSE
  test_cases.csv
  contigs.csv
  contig_specifications.csv
  motif_cases_metadata.csv
  cluster_membership.csv
  motif_pdbs/
  original_pdbs/

Files

• —test_cases.csv: MotifBench-style case table with benchmark PDB identifiers, motif residues, redesignable indices, total length, group label, and non-motif context.
• —contigs.csv: Contig strings for scaffold-generation workflows using original residue numbering.
• —contig_specifications.csv: Normalized and original contig specifications for each benchmark problem.
• —motif_cases_metadata.csv: Extended metadata for each case, including PROSITE entry, motif name, pattern, source PDB/chain, motif segments, motif length, redesignable and non-motif regions, assembly annotation, structural-conservation statistics, and multi-conformational annotation.
• —cluster_membership.csv: Mapping between benchmark representatives, case identifiers, PROSITE entries, and cluster membership roles.
• —motif_pdbs/: Motif-only PDB files for each exported benchmark case.
• —original_pdbs/: Corresponding source-structure PDB files used for benchmark export.

Dataset summary

The dataset contains:

• —220 benchmark cases
• —174 unique PROSITE motif-pattern entries
• —median motif length of 19 amino acids
• —assembly-context annotations: core, ribosomal, and huge_assembly

The benchmark is focused on within-chain, structurally conserved functional motif signatures. Because PROSITE patterns are primarily linear or near-linear sequence signatures, many cases are compact and contiguous or near-contiguous. Multi-segment cases arise when long internal wildcard spans are preserved as non-motif scaffold context.

Intended use

This artifact is intended to support reproducible evaluation of protein motif-scaffolding models. A typical workflow is:

1. 1.Read case definitions from test_cases.csv and/or contigs.csv.
2. 2.Use the corresponding motif-only PDB files in motif_pdbs/ as fixed motif inputs.
3. 3.Generate scaffolds under a fixed sampling budget, for example 100 generation attempts per case.
4. 4.Evaluate outputs using a MotifBench-style refolding and motif-recovery pipeline.
5. 5.Use metadata in motif_cases_metadata.csv for stratified analyses by motif class, assembly context, motif length, structural-conservation statistics, or multi-conformational annotation.

Reproducibility notes

• —Residue labels in exported benchmark cases are normalized for compatibility with motif-scaffolding pipelines.
• —Original residue labels and source-structure information are preserved in the metadata tables.
• —The original PDB structures and PROSITE documentation are public upstream resources; this archive repackages selected case definitions and derived annotations for benchmark evaluation.

Anonymity note

This artifact is prepared for anonymous peer review. Author names, affiliations, and institution-specific paths have intentionally been omitted.

License and attribution

The benchmark case definitions, metadata, and added annotations in this archive are released under the terms in LICENSE.

The archive contains data derived from public upstream resources, including PROSITE and PDB structures. Users should cite and comply with the terms of the upstream resources where applicable.