AdaptivePy

AdaptivePy performs adaptive sampling on molecular dynamics trajectories using clustering-based state space partitioning and policy-driven seed selection. It also supports frame-level policies such as MaxEnt VAMPNet and TS-DAR, which score individual frames without clustering.

Install

pip install adaptivepy-sampling

For Torch-backed policies:

pip install adaptivepy-sampling[torch]

Quick links

• Getting Started — run your first analysis in minutes
• Configuration — YAML options and defaults
• Feature Inputs — supported file formats and layout
• CLI — command-line usage
• Python API — programmatic access
• API Reference — full module documentation

What it does

1. Load per-trajectory feature arrays (.npy or .pkl)
2. Cluster frames in feature space (skipped when only frame-level policies are used)
3. Apply one or more adaptive policies (least_counts, random, fast, ma_reap, knn_as, maxent_vampnet, ts_dar)
4. Select seed frames from chosen clusters or directly by frame-level scores
5. Write metadata, assignments, policy-specific scores, and optional PDB structures

Example

adaptivepy run examples/config.yaml

See the Getting Started guide for a complete walkthrough.