VERSION HISTORY

Changelog

A record of engine updates, validation milestones, and platform changes. The engine version is frozen at v5.1 — all changes below this point are platform and validation updates only.

v5.1.4validationApril 5, 2026

p53 Y220C Benchmark & Demo Portfolio Finalized

  • •Added p53 Y220C (6SHZ apo + 9BR4 ligand-bound) as rescued-pocket benchmark — 93/100 apo, 83/100 ligand-bound, same 338 Da primary lead in both states
  • •Completed β-catenin 5-structure conformational series with 1QZ7 (Axin complex) — 83/100, 312 Da, Strong coherence
  • •Demo portfolio finalized: 7 cases showing 3 discrimination outcomes (yes-but-carefully / no-clearly / this-domain-not-that-one)
  • •Added trust and policy pages: Privacy Policy, Terms of Use, Security Overview, Limitations, Changelog
  • •Rebuilt About Us and Homepage for investor focus — 3-outcome discrimination story
  • •Trimmed public navigation to 5 essential items
v5.1.3validationApril 4, 2026

MYC & α-Synuclein Hard Negatives

  • •Added MYC (1NKP) as undruggable benchmark — 12/100, 0/100 top-pocket, non-small-molecule constrained
  • •Added α-synuclein (1XQ8) as undruggable benchmark — 3/100, 0/100 top-pocket
  • •Demo Cases page launched with flagship β-catenin comparison table
  • •Strategy badge mapping fixed for non_small_molecule_constrained classification
v5.1.2engineApril 3, 2026

Layer F Molecular Hypothesis Engine

  • •Deployed Layer F — molecular hypothesis generation from pocket geometry and LPD field
  • •Generates molecular weight, formula, mechanism class, and SMILES hypotheses
  • •Coherence scoring validates consistency across top pockets
  • •Strategy classification: small_molecule, biologic_preferred, non_small_molecule_constrained
v5.1.1platformApril 2, 2026

Multi-State Analysis & Conformation Compare

  • •β-catenin analyzed across 4 states: AlphaFold predicted, BCL9/TCF4 (2GL7), hTcf-4 (1JDH), Compound 6 (7AFW)
  • •Conformation comparison tool launched — side-by-side structural state analysis
  • •TDP-43 domain prioritization: RRM1 identified as strongest small-molecule-compatible region
v5.1.0engineApril 1, 2026

V5 Engine Frozen

  • •Engine v5.1 frozen with 10 proprietary atomic interaction constants
  • •Six-layer decision architecture (A through F) finalized
  • •LPD field computation, sigmoid scoring, binding site discovery (peaks + valleys)
  • •Based on new physics concept that also solved the many-body problem
v5.0.0engineMarch 2026

V5 Engine Development

  • •Complete rewrite of the Ashebo engine from first principles
  • •New atomic interaction constants derived from foundational physics
  • •Blind validation against known druggable and undruggable targets
  • •89% accuracy on blind benchmark set (p < 0.05)