AlphaFold Substrate Alignments#

Scope and Assumptions#

1. Scope#

1.1 Systems Covered#

This substrate applies to biological inference engines that exhibit the following characteristics:

• high‑dimensional latent representations of protein structure
• iterative or multi‑stage inference cycles
• attention‑based or pairwise‑embedding architectures
• structural outputs expressed as 3D conformations or residue‑level predictions
• regime‑dependent folding coherence signals

The primary reference system is AlphaFold, but the substrate is compatible with any model exhibiting similar inference behavior.

1.2 Structural Focus#

The substrate addresses:

• folding‑pathway coherence
• motif‑level structural stability
• latent‑space orientation and projection
• regime transitions during inference
• dimensional‑core alignment (3D–9D)

It does not address biochemical energetics, molecular dynamics simulations, or experimental folding mechanisms.

1.3 Intended Use Cases#

The substrate supports:

• interpretability of protein‑folding inference models
• cross‑model comparison and alignment
• drift detection in biological inference systems
• reproducibility analysis
• dimensional projection and regime mapping
• integration with vST validation layers

2. Assumptions#

2.1 Model Behavior Assumptions#

The substrate assumes:

• the inference system produces stable latent‑space structures
• folding predictions converge toward coherent surfaces
• regime transitions follow identifiable resonance‑time patterns
• dimensional projections preserve structural invariants
• inference cycles are deterministic or quasi‑deterministic under fixed inputs

These assumptions reflect observed behavior in AlphaFold‑class systems.

2.2 Data and Input Assumptions#

The substrate assumes:

• input sequences are fixed and pre‑validated
• multiple sequence alignments (MSAs) or equivalent features are available
• structural outputs are expressed in 3D coordinate form
• inference noise is bounded and does not dominate regime transitions

The substrate does not require access to training data or model internals.

2.3 Biological Assumptions#

The substrate assumes:

• protein structures exhibit stable motif‑level coherence
• folding pathways can be represented within 3D–9D dimensional cores
• biological variability does not invalidate substrate‑level invariants

The substrate does not assume any specific biochemical mechanism.

3. Out‑of‑Scope Elements#

The following are explicitly out of scope:

• molecular dynamics simulations
• thermodynamic or kinetic modeling
• experimental structure determination
• biochemical pathway analysis
• evolutionary modeling beyond MSA‑derived features

These domains may interface with the substrate but are not defined by it.

4. Validity Conditions#

Substrate alignment is valid when:

• inference outputs are structurally coherent
• latent‑space representations are stable across inference cycles
• dimensional projections preserve motif‑level structure
• regime transitions follow triadic resonance patterns
• vST validation layers confirm reproducibility

If these conditions are not met, substrate interpretation may be incomplete.

5. Dependencies#

This document depends on:

• substrate_definition.md for axis and primitive definitions
• alignment_principles.md for mapping rules
• dimensional_cores.md for 3D–9D substrate structure
• validation_layers_vst.md for reproducibility and drift checks