인터페이스는 현지화되어 있으며, Skill 소개와 설명 문서는 선택한 언어로 AI가 자동 번역합니다. 코드·명령어·경로·전문 용어는 원문을 유지하며 원작자 문서가 우선합니다.
Skills 둘러보기
데이터 및 연구 · SKILL.md

selections-topology

selections-topology

Use and debug MDAnalysis atom selections, topology attributes, groups, fragments, topology guessing, and selection exporters.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill 문서

GitHub

name: selections-topology description: "Use and debug MDAnalysis atom selections, topology attributes, groups, fragments, topology guessing, and selection exporters." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

MDAnalysis Selections and Topology

Use this sub-skill when a task involves select_atoms(), AtomGroup/ResidueGroup/SegmentGroup, topology attributes, group or updating selections, bonds/fragments, topology guessing, or exporting selections for another molecular tool.

Read First

  • Start with Selection Language for selection grammar, keyword behavior, sorting, dynamic selections, selection groups, geometry, SMARTS, and exporters.
  • Use Topology and Groups for group operations, topology attribute access, custom attributes, fragments/bonds, and topology guessing.
  • Use Troubleshooting when selections are empty, reordered, slow, dependency-sensitive, or raise SelectionError, NoDataError, ValueError, or optional RDKit-related errors.
  • Run python scripts/selection_probe.py from this sub-skill directory to confirm the installed MDAnalysis selection/topology basics with only synthetic data.

Common Task Routes

  • Selection text: Prefer u.select_atoms("...") or ag.select_atoms("..."); remember keywords and string values are case-sensitive and selections are unique/sorted by topology index unless sorted=False is passed.
  • Preserve order: For manually ordered groups such as ag[[5, 1, 0]], pass sorted=False to select_atoms() or use group concatenation/slicing when duplicates must be preserved.
  • Dynamic selections: Pass updating=True for an UpdatingAtomGroup that lazily re-evaluates after trajectory frame changes; avoid it for static topology-only selections in tight loops.
  • Topology attributes: Add built-in attributes with u.add_TopologyAttr("masses", values) or custom TopologyAttr subclasses; selectable custom attributes need supported dtype values (bool, integer, float, string/object).
  • Cross-skill routing: Send file loading/writing format decisions to ../universe-io/SKILL.md, converter interoperability to ../formats-converters/SKILL.md, and numerical distance/analysis algorithms to ../analysis-workflows/SKILL.md.
저장한 Skills