SQQ

SQQ (Shell Quant Qualifier): Python Joint Toolkit for Water-Shell Topology Analysis.

Current development version: 0.5.7 (Sep 25, 2026; Mid-Autumn Day)

SQQ identifies water-network rings, cages, hydrate phases, guest occupancy, order parameters, and persistent cage tracks from molecular-dynamics structures and trajectories. It provides the complete SQQ-Py workflow and a focused native SQQ-CPP cage engine.

Detailed definitions are in docs/design.md; release history is in docs/update.md.

Changed in 0.5.7

• Reduced memory use for long trajectories and large Track runs.
• Faster render output, imported-result validation, and tracking of selected cages.
• Clearer progress and warnings, with unchanged scientific results for the same input.

See Version 0.5.7 for complete release notes.

Acknowledgements

Names are listed alphabetically by family name.

• Cao, Pinqiang @ Wuhan University of Science and Technology
• Cheng, Liwei @ Wuhan Institute of Technology
• Fang, Bin @ Hainan University
• Hu, Yifei @ Fuzhou University
• Jia, Jihui @ China University of Petroleum (Beijing)
• Li, Wuquan @ Beijing Huairou Laboratory
• Li, Zhenchao @ Fuzhou University
• Liao, Bo @ China University of Petroleum (East China)
• Lu, Yingxu @ Wuhan Institute of Technology
• Mi, Fengyi @ Southwest University of Science and Technology
• Sun, Yingtao @ The Hong Kong University of Science and Technology
• Xu, Hongye @ The University of Tokyo
• Zhang, Zhengcai @ Laoshan Laboratory
• Zhao, Jingyuan @ Akatsuki Games Inc.

Installation & Update

pip install sqq
pip install --update sqq

For local development:

pip install -e .
python -m sqq -h

Source builds require Python 3.10+, CMake 3.20+, a C++17 compiler, Python development headers, and a platform build tool. Released wheels normally include the native extension. A first invocation may take a few extra seconds while dependencies create a Matplotlib font cache.

sqq --version
sqq -h

Quick Start

The default py engine analyzes the water network, 4/5/6-membered rings and cages, F3/F4, writes per-frame information, creates a VMD package, and produces summary.xlsx.

Configuration

sqq init
sqq analyze -i test.gro -c sqq_config.yaml -o ./result_sqq

sqq init writes a commented sqq_config.yaml and refuses to overwrite it. SQQ does not auto-load this file: use -c explicitly, or omit -c to use built-in defaults.

Structures and trajectories

# One GRO structure
sqq analyze -i test.gro -o ./result_sqq

# A directory or glob of structures
sqq analyze -i ./gro -o ./result_sqq
sqq analyze -i "./gro/*.gro" -o ./result_sqq

# XTC/TRR plus topology; sample every 100 ps
sqq analyze -i traj.xtc -t topol.gro -dt 100 -o ./result_sqq

# Repeated complete GRO frames
sqq analyze -i frames.gro -dt 100 -o ./result_sqq

# LAMMPS dump/DCD plus DATA topology
sqq analyze -i traj.lammpstrj -t system.data -o ./result_sqq

Without -dt, every stored frame is analyzed. Compatible GRO files share one result root; incompatible topologies are separated into result_A, result_B, and so on.

Engines, workers, and explicit pairs

sqq analyze -i ./gro -e py -w 4 -o ./result_py
sqq analyze -i traj.xtc -t topol.gro -e cpp -w 50% -o ./result_cpp
sqq analyze -i test.gro -b pairs --pair water_pairs.txt -o ./result_pairs

Use py for the complete workflow and cpp for the focused native cage workflow. Integer worker values are counts; decimals and percentages are physical-core fractions.

Track cages

# Reuse an Analyze result
sqq track --source ./result_sqq --target all -o ./result_track

# Or track directly from a trajectory
sqq track -i traj.xtc -t topol.gro -dt 100 --target 512,t133 -o ./result_track

Track targets may be all, cage types, hydrate phases, or persistent IDs such as t133.

Visualize results

# Validate render files and print absolute VMD commands
sqq vmd ./result_sqq

Engines and Inputs

Values 00 and 99 are compatibility presets, not points on a continuous scale. The default is py.

SQQ-Py provides the complete graph/ring/half/quasi/cage/cluster/phase/ice/order-parameter workflow. SQQ-CPP provides native graph construction, internal chordless 4/5/6 rings, cage/isomer/occupancy, and F3/F4. Native errors never silently fall back to Python.

Only orthorhombic periodic boxes are supported. LAMMPS can infer common water and methane components; explicit type/role mappings take priority. Walls, surfactants, additives, and other retained components do not enter the water graph unless classified as water. See Input Validation and Coordinate Units.

CLI Reference

--pair implies pairs mode unless -b pairs is already present; it conflicts with an explicit non-pairs mode. CLI-relative pair paths resolve from the working directory, while YAML-relative paths resolve from the configuration directory. Expected user-facing failures print one concise Error: ...; SQQ_DEBUG=1 enables development tracebacks. Full CLI and migration rules are in Analysis Engines and Workers.

Configuration

Precedence is:

built-in defaults -> engine preset -> sqq_config.yaml -> retained CLI overrides

A compact configuration is:

engine: py
water:
  resname: [SOL, TIP, WAT, HOH]
guest:
  resname: [CH4, CO2, MET, ETH]
additive:
  resname: []
environment:
  resname: []
graph:
  mode: auto
ring:
  size: [4, 5, 6]
  report_size: auto
half_cage:
  enabled: auto
quasi_cage:
  enabled: auto
  max_layer: 1
cage:
  report_type: auto
  scientific_validation: false
hydrate_cluster:
  enabled: false
order_parameter:
  enabled: [f3, f4]
output:
  type: [info, sqq-render, summary-xlsx]
render:
  atom_scope: full
track:
  target: all
  gap_frame: 0
  max_gap_ps: null

The generated YAML contains every key, default, unit, and choice. Unknown or duplicate keys are errors; supported older spellings migrate with a warning. Every run writes effective settings, provenance, failures, timing, and outputs to sqq_config_resolved.yaml. See the complete Configuration Reference.

Scientific Scope

SQQ builds an hbond, oo, or explicit-pair water graph; finds rings and half/quasi cages; validates closed-cage topology; assigns guest occupancy; and optionally classifies hydrate domains, ice-like water, F3/F4, Q_l, MCG, and DHOP. Search and report scopes are separate. Diagnostic cage-state limits abort a frame instead of publishing partial cage results. Optional geometry validation may remove distorted cages and consequently change occupancy or cluster output. Exact definitions are in docs/design.md.

Tracking

sqq track assigns persistent t1, t2, ... IDs from stable water identities and deterministic cage matching.

Mixed targets are written independently, and type/phase targets retain complete selected lifecycles. Phase targets require SQQ-Py cluster labels; imported state cannot create missing labels retroactively. A target that matches no cage is still written (empty tables, render package without selections) and is named in the final Warnings row.

Matching remains water-shell-first; guest exchange does not create a cage ID. Each continuation is checked against the track's most recently observed shell, so gradual water exchange can preserve an ID while every adjacent or gap-bounding pair still passes the configured thresholds even after overlap with the initial shell reaches zero. Version 0.5.6 diagnostics and derived statistics do not change the default tID assignment. Raw Track is currently serial and is required for pre-cage precursor history; since 0.5.7 that history is reconstructed from compact per-frame records spooled during the single analysis pass, so the prefix is not analyzed twice.

Raw Track validates the final effective time of every selected frame before analysis, including configured fallback times for frames without stored time. Times may be irregular or equal but cannot decrease; physical gaps and duration statistics use those actual times, while gap_frame remains a selected-frame count. Analyze keeps valid per-frame scientific output if snapshot reduction is unsuitable, but omits that topology group's persistent Track state with a warning. Schema 2/3 state can recover partial diagnostics from archived thresholds; schema 1 retains its raw evidence without inventing threshold-dependent classifications.

Only directly consecutive observations produce resolved type, phase, guest, or occupancy changes. A change across a permitted recognition gap is recorded as unresolved, and affected residence segments remain gap-censored lower bounds. Split/merge confirmation additionally requires persistent destination shells and independent branch-water contributions; overlap between already persistent neighboring cages is retained only as non-confirmable candidate evidence.

result_track/track/
  track_state.json
  {cage_observation,cage_track,cage_event,cage_population}.csv
  {guest_residence,lifetime_distribution}.csv
  statistics/{tracking_quality,lifetime_survival,cage_lineage,...}.csv
  network/{cage_transition_nodes,cage_transition_edges}.csv
  <target>/
    filtered tables and statistics
    precursor_state.csv, water_history.csv       # persistent-ID target; source may report unavailable
    sqq_render/{sqq_track.gro,sqq_track.xtc,
                sqq_track.membership.tsv,sqq_track.vmd.tcl}

See Cross-Frame Cage Tracking for exact matching, duration, event, network, and source/raw definitions.

Track target GRO/XTC files may be hard-linked to avoid duplicate trajectory data. Treat them as immutable SQQ output: deleting or atomically replacing one path is safe, but editing a hard-linked file in place can also alter a sibling target package.

New Analyze state binds its matching configuration and complete render package to topology, component, atom, cage, selected-frame/time, and SHA-256 provenance; the digests are taken while the package is written. sqq track --source re-hashes and validates those records before publishing any target result; older states remain readable through structural validation.

Track workflow memory is bounded by active/dormant matching state rather than the complete observation history: observations and events use a run-private disk spool, the JSON state and observation-level tables are streamed record by record, and per-target tables reuse the run-level rows. The public in-memory tracking API remains available for programmatic use. The release gate includes a real 1.417 GB / 1,001-frame LAMMPS run plus exact SQQ-Py/SQQ-CPP Track-output comparisons; no scientific tolerance is widened for these optimizations.

Outputs

default,summary-detail-csv extends the engine default. all and none are exclusive. Track output is fixed and cannot be reduced with --output-type.

result_sqq/
  sqq_config_resolved.yaml
  sqq_output_manifest.json
  summary.xlsx and/or summary/
  info/frame_*_info.md
  gro/<frame>/                         # selected structures
  sqq_render/
    sqq_cage.gro
    sqq_cage.xtc
    sqq_cage.membership.tsv
    sqq_cage.vmd.tcl
  track/
    track_state.json
    compatibility tables
    statistics/
    network/

A single GRO/XYZ may use a compact frame-root layout. Incompatible topologies use complete result_A through result_Z roots; larger heterogeneous sets use an information-only safety layout. Main summaries contain the applicable graph, ring, cage, cluster, ice, and order tables; selected detail output adds occupancy and isomer tables. SQQ-CPP writes its applicable subset. With all, cluster outputs still require cluster search.

All successful Analyze sequences write persistent Track state independently of sqq-render. If a requested output directory is nonempty, SQQ preserves it and selects _001, _002, and so on. Outputs are staged and atomically published. See Output Layout for complete files and schemas.

VMD

Keep all four sqq_render/ files together. Use sqq vmd RESULT to validate them and print absolute commands, or source the Tcl script directly:

source {path/to/result_sqq/sqq_render/sqq_cage.vmd.tcl}

sqq show <family> <target...> [<family> <target...>]...
sqq color <family> <target...> <VMD-color|ColorID|default>
sqq show label [on|off]
sqq pick center|guest|off
sqq target save
sqq clear
sqq help | sqq -h | sqq --help

Families are cage, guest, phase, cluster, domain, and component. For example:

sqq show cage 512 guest 512
sqq show phase sI component environment
sqq color cage 512 green
sqq pick center
sqq target save

The default is opaque cage all with context hidden. Picking makes unselected objects transparent, highlights cages in yellow and guests in orange, and leaves labels off. DynamicBonds style changes persist across earlier and later frames; explicit SQQ style/reset commands may intentionally replace them. Full picking, color, target-validation, and representation rules are in Output Layout.

Terminal and Citation

Interactive runs update one live progress region and then redraw a compact final page with basic information, resolved configuration, frame/timing totals, optional VMD commands, and:

Cages were identified and analyzed using SQQ.
Publication: J. PANG & Q. SUN. SQQ: Python Joint Toolkit for Water-Shell Topology Analysis. Journal of Physical Chemistry A.
DOI: http://doi.org/10.1021/acs.jpca.6c05769
GitHub: https://github.com/pimooni/sqq

Redirected output remains plain append-only text.

Python API

from sqq import analyze_frame, load_config, read_frames

config = load_config({"graph": {"mode": "oo"}}, engine="py")
frame = next(read_frames("frame.gro", config=config))
result = analyze_frame(frame, config)

These public functions use the same readers, configuration resolution, validation, and engines as the CLI. They return immutable resolved configuration, Frame, and FrameResult objects and raise typed SQQ exceptions. sqq.api and sqq.models are the stable import surfaces; workflow, runtime, and I/O module paths are implementation details and may move between releases. Track record classes in sqq.models are slotted immutable dataclasses: use dataclasses.fields/dataclasses.asdict for generic inspection rather than vars() or weak references. Their name-based pickle state remains readable from 0.5.6.

Current Limits

• Periodic geometry is orthorhombic; triclinic input must be converted first.
• XYZ has configurable scaling but no periodic box.
• Closed cages use 4/5/6 faces; 7-member rings remain available to ring/quasi analysis.
• SQQ-CPP omits public ring, half/quasi, cluster/phase, ice, Q_l/MCG/DHOP, and related category output.
• Domain/cluster IDs are frame-local; Track assigns persistent IDs to cages, not grains.
• Source Track cannot reconstruct pre-cage history; use raw Track when required. A Track result is not itself a valid --source.
• Raw Track accepts one physical trajectory or stacked GRO system; select incompatible Analyze groups separately.
• Exit status is 0 for a completed run (even with some failed non-strict frames), 1 when every requested frame failed, and 2 for configuration, input, or runtime errors.

Documentation

• Complete design and reference
• Release notes
• sqq analyze -h — Analyze options
• sqq track -h — Track targets and inputs
• sqq vmd -h — VMD locator and Tcl commands
• sqq init — complete commented configuration