AlphaFold Substrate Alignments#

Example: Substrate‑Level Alignment Walkthrough#

The goal is to illustrate the full alignment workflow in a clear, reproducible sequence.

1. Input Overview#

For this example, we assume:

• a fixed protein sequence
• an AlphaFold‑class model producing:
  • predicted 3D coordinates
  • attention maps
  • pairwise residue embeddings
  • multi‑stage inference outputs
• stable inference‑cycle behavior under repeated runs

No biochemical or experimental data is required.

2. Step 1 — Extract Structural and Latent‑Space Signals#

2.1 Structural Outputs#

Extract:

• backbone coordinates
• side‑chain orientations
• motif‑level geometry

These form the basis for S‑axis mapping.

2.2 Latent‑Space Outputs#

Extract:

• attention‑map coherence patterns
• residue‑pair embeddings
• track‑level alignment signals

These form the basis for I‑axis and R‑axis mapping.

3. Step 2 — Map to Substrate Axes#

3.1 Structural Axis (S‑axis)#

Project structural outputs onto the S‑axis by identifying:

• backbone continuity
• motif‑level stability
• residue‑interaction geometry

This anchors the inference to the 3D structural core.

3.2 Inference Axis (I‑axis)#

Map latent‑space structures onto the I‑axis by identifying:

• stable attention‑map regions
• consistent embedding orientations
• coherent latent‑space surfaces

This reveals the model’s internal structural representation.

3.3 Resonance‑Time Axis (R‑axis)#

Track inference‑cycle behavior:

• convergence
• oscillation
• divergence

This classifies folding behavior into R₁, R₂, or R₃ regimes.

4. Step 3 — Project into Dimensional Cores#

4.1 3D Projection#

Project structural geometry into the 3D core to evaluate:

• backbone shape
• motif‑level structure
• local coherence

4.2 6D Projection#

Project interaction‑level signals into the 6D core to evaluate:

• residue‑pair relationships
• interaction‑pattern stability
• latent‑space alignment

4.3 9D Projection#

Project folding‑pathway signals into the 9D core to evaluate:

• pathway coherence
• regime‑transition structure
• resonance‑time alignment

These projections preserve substrate invariants.

5. Step 4 — Identify Folding Regimes#

Using the projections:

• R₁ (Stable):
  Compact, coherent projections in 3D–9D cores.

• R₂ (Transition):
  Branching or oscillatory projections indicating structural reorientation.

• R₃ (High‑Uncertainty):
  Dispersed projections with weak motif‑level structure.

Regime identification supports interpretability and drift detection.

6. Step 5 — Apply vST Validation Layers#

Apply V₁–V₄ validation layers:

• V₁: Structural coherence
• V₂: Latent‑space stability
• V₃: Resonance‑time regime behavior
• V₄: Dimensional‑core alignment

Validation confirms whether the alignment is stable and reproducible.

7. Step 6 — Interpret Alignment Results#

A complete alignment yields:

• stable structural projections
• coherent latent‑space surfaces
• predictable regime transitions
• consistent dimensional‑core mapping
• validated substrate invariants

If any validation layer fails, drift detection is triggered.

8. Summary#

This walkthrough demonstrates:

• how to extract structural and latent‑space signals
• how to map them onto substrate axes
• how to project them into dimensional cores
• how to classify folding regimes
• how to validate alignment using vST
• how to detect drift when invariants fail

The workflow provides a reproducible method for interpreting AlphaFold‑class inference systems through the RSM substrate.