Skip to main content

SEAMM Extract Clusters Plug-in

GitHub pull requests Build Status Code Coverage Code Quality Documentation Status Updates for Dependencies PyPi VERSION

A SEAMM plug-in for extracting molecular clusters (trimers, tetramers, … larger n-mers) from a condensed-phase, typically periodic, configuration as unwrapped, non-periodic structures – e.g. many-body training data and diagnostics for machine-learned force fields.

Features

  • Clusters are connected subgraphs of a molecular contact graph (molecules are in contact if their contact atoms are within a cutoff, minimum image), so chains, rings and stars all occur – not just the most compact cluster.

  • Any cluster size, and several sizes per frame (e.g. 3, 4).

  • Works on many frames in one step – all configurations of a system holding a trajectory, or any standard SEAMM structure selection – so no loop is needed and the clusters land in one system.

  • Stratification so the set is flat in a spread coordinate (radius of gyration or largest centroid separation), with bin edges from equal quantiles of a pilot sample or given explicitly; optional balancing over the contact-graph motif.

  • Clusters are unwrapped across the periodic boundary, centred, non-periodic, with molecules and bonds intact.

  • Provenance on every configuration: unique names <frame>_<seed>_<m1-m2-...> and #ExtractClusters#scan properties (size, spread, motif, contacts, source molecules) that survive SDF/extxyz export; a clusters.csv summary per step.

  • Works on any cell (orthorhombic fast path via a periodic KD-tree; exact minimum image otherwise) and on non-periodic sources.

Acknowledgements

This package was created with Cookiecutter and the molssi-seamm/cookiecutter-seamm-plugin project template.

Developed by the Molecular Sciences Software Institute (MolSSI), which receives funding from the National Science Foundation under award ACI-1547580

History

2026.9.18.1 – Extract from many structures in one step
  • The step now takes the standard SEAMM structure selection: the current configuration (the default, as before), all or the last or first configurations of the current system, of all systems, or of systems chosen by name, or a variable holding a list of configurations. Selecting all the configurations of a system that holds a trajectory extracts from every frame in one step, with no loop, and puts all the clusters in one system. Each frame’s clusters are prefixed with its configuration name; clusters.csv and summary.json record the frame; the report aggregates over frames and lists the count per frame. One random stream covers the whole run, so the printed seed reproduces it.

  • Requires seamm 2026.9.18.1 and molsystem 2026.9.17.2 or later.

2026.9.18 – Reproducible seeds and motif selection
  • The random seed actually used is now always printed and recorded in a new summary.json in the step directory (with the source, bin edges and counts by motif), so a run made with the seed set to “random” can be reproduced by entering the printed value.

  • A new “Restrict to motifs” option accepts only clusters with the given contact-graph topology, e.g. ring or ring, star. The dialog offers the motifs the requested cluster sizes can produce. Candidates rejected for their motif are counted and reported, since rare motifs use up the attempt budget.

2026.9.17.1 – Bugfix: a structure without bonds gave clusters of atoms
  • Molecules are identified from the bonds, so a configuration read from a format that carries no connectivity (extended XYZ without bond perception) was treated as one atom per molecule and the “clusters” were silently groups of atoms. The step now stops with a clear error pointing at the Read Structure “Perceive bonds” option. Configurations made only of noble-gas atoms or monatomic ions, which legitimately have no bonds, are still accepted.

2026.9.17 – Initial release of the Extract Clusters step
  • Extracts n-molecule clusters (trimers, tetramers, … larger n-mers) from the current, typically periodic, condensed-phase configuration as unwrapped, centred, non-periodic configurations in a new system, e.g. many-body training data and diagnostics for machine-learned force fields.

  • Clusters are connected subgraphs of a molecular contact graph (molecules are in contact if their contact atoms are within a cutoff, minimum image), so chains, rings and stars all occur; several sizes can be extracted from one frame.

  • Optional stratification so the set is flat in the radius of gyration or the largest centroid separation, with bin edges from equal quantiles of a pilot sample or given explicitly, and optional balancing over the contact-graph motif.

  • Provenance on every cluster: unique names <frame>_<seed>_<molecules> and #ExtractClusters#scan properties (size, spread, motif, contacts, bin, source molecules) that survive SDF/extxyz export, plus a clusters.csv per step.

Metadata

Release files for extract-clusters-step 2026.9.18.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for extract-clusters-step 2026.9.18.1
File Size Uploaded
extract_clusters_step-2026.9.18.1.tar.gz 72.7 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for extract-clusters-step 2026.9.18.1
File Interpreter ABI Platform
extract_clusters_step-2026.9.18.1-py2.py3-none-any.whl Python 3, Python 2 none any Details

Total release size: 106.7 kB

Release files / extract_clusters_step-2026.9.18.1.tar.gz

Download URL extract_clusters_step-2026.9.18.1.tar.gz
Size 72.7 kB
Tags Source
SHA-256 checksum
How to use checksums
136a441ee528c3bbc8160285df614ae1911f1b444b556a0e4129198480126c57
BLAKE2b-256 checksum
How to use checksums
c484abf41d70d90bd1055cf8236407a20f7b2e70d4e0c1d93614715894794f6c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14

Release files / extract_clusters_step-2026.9.18.1-py2.py3-none-any.whl

Download URL extract_clusters_step-2026.9.18.1-py2.py3-none-any.whl
Size 34.0 kB
Tags Python 2 Python 3
SHA-256 checksum
How to use checksums
af2b5c2e476f93aa88765dd1f3474ea16585cf01812c409fed632a739a9c5898
BLAKE2b-256 checksum
How to use checksums
eab97be12a53ad4c68efefb20942d839ed7fa649cd55079ad2ea0c32f2fc76d5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14

Release history Release notifications | RSS feed

This release

2026.9.18.1 This release

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page