input-data-and-features
Author, validate, and convert Protenix input JSON for proteins, nucleic acids, ligands, ions, paths, bonds, and constraints without running inference.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- input-data-and-features
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Author, validate, and convert Protenix input JSON for proteins, nucleic acids, ligands, ions, paths, bonds, and constraints without running inference.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/conversion-and-features.md7.1 KB
- 参考文档references/input-json-schema.md9.0 KB
- 参考文档references/troubleshooting.md7.9 KB
- 脚本scripts/protenix_minimal_input.json199 B
- 脚本scripts/validate_protenix_input_json.py10.5 KB
03 SKILL.md 正文
Input Data and Features
Use this sub-skill when a task is about preparing Protenix input before prediction: authoring JSON jobs, validating entity blocks, choosing ligand encodings, checking MSA/template path fields, defining covalent bonds or constraints, or planning structure-to-JSON conversion.
Use This For
- Build the top-level list-of-jobs JSON with
name,sequences, optionalcovalent_bonds, optionalconstraint, and optionalmodelSeeds. - Author
proteinChain,dnaSequence,rnaSequence,ligand, andionentries with correctcount, optionalid, modifications, and sequence alphabets. - Fill existing
pairedMsaPath,unpairedMsaPath, andtemplatesPathfields without running MSA/template searches. - Encode ligand
CCD_,SMILES, andFILE_forms, bare-code ions, polymer-ligand bonds, ligand-ligand bonds, contact constraints, and pocket constraints. - Plan
protenix jsonconversion from PDB/mmCIF/CIF or use conversion API signatures when an installed Protenix environment is available.
Route Elsewhere
- Run prediction, choose model variants, tune inference flags, or interpret output confidence files with
../cli-and-inference/SKILL.md. - Generate missing protein MSA, template, or RNA MSA files with
../msa-template-and-prep/SKILL.md. - Prepare training datasets, bioassembly caches, or training-time data schemas with
../training-and-data-pipeline/SKILL.md.
Start Here
- Read
references/input-json-schema.mdfor accepted job/entity/path/bond/constraint shapes and copyable snippets. - Run
python scripts/validate_protenix_input_json.py scripts/protenix_minimal_input.jsonfor a validator smoke test. - Run
python scripts/validate_protenix_input_json.py INPUT.json --check-pathsbefore handing authored JSON to prediction or preprocessing. - Read
references/conversion-and-features.mdforprotenix json,cif_to_input_json, and feature-conversion behavior. - Read
references/troubleshooting.mdwhen JSON is rejected, paths are missing, chemistry parsing fails, or conversion output looks surprising.
Critical Reminders
- The top-level JSON value is always a list, even for one job.
- Entity numbers in bonds and constraints are 1-based positions in
sequences, not chain IDs. - Ligand values use
CCD_,FILE_, or SMILES; ion values use bare codes such asMGorZN. pairedMsaPath,unpairedMsaPath, andtemplatesPathshould point to existing files, preferably by absolute path.protenix jsonconverts structures to input JSON; it does not run prediction or generate missing MSA/template files.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).