Rose Engine — Dynamic Biological Systems Interaction Explorer

Rose Engine is a computational biology research platform investigating how synonymous codon choice, RNA modification, metabolic context, immune state, and structural biology jointly shape disease and treatment response across 52 diseases. Every result is grounded in 22 published biochemical databases with complete evidence chains and no synthetic data.

Operated by Rose Engine, LLC. Live site: https://roseengine.io/.

What the platform does

Builds patient-specific treatment plans with Hill-equation dosing and CYP / renal / hepatic adjustment.
Generates cure blueprints that grade every proposed target level-by-level (Mechanism → Fold → Pocket) with a CAN / CANNOT structural-designability verdict from real AlphaFold pLDDT, binding-pocket residues, and ChEMBL bioactivity — ungrounded designs are reported as insufficient data, never speculative coverage.
Surfaces the PDB accession and literature citation behind every binding-pocket residue in the Designability panel for reviewer verification; exports a downloadable structure-only evidence list and honest data-gap audit sheet alongside the Wet-Lab Handoff Sheet (CSV / HTML / JSON).
Provides a searchable and filterable Designability list across all enzymes and gap types in the Cure Blueprint.
Routes failed small-molecule targets through seven therapeutic modalities — siRNA, ASO, CRISPR knockout, mRNA replacement, and base / prime / ADAR editing — emitting a grounded construct only when a real RefSeq CDS or ClinVar variant resolves.
Attaches per-construct in-silico validation readiness chains (efficacy, off-target, safety, delivery, dose, causal) and exports a downloadable wet-lab handoff sheet (CSV / HTML / JSON) for every designed construct.
Designs RNA therapeutics (siRNA, ASO, mRNA, ADAR guide) and ribosome-targeted therapeutics ranked using five wired ribosome data layers.
Computes ViennaRNA RNA secondary-structure folds for every rRNA modification site (41-nt windows), reporting local minimum free energy and per-position unpaired probabilities.
Integrates the ribosome PDB structure (4UG0) with 3D distance annotations — per-site distances to the peptidyl-transferase center, decoding center, and exit tunnel, plus proximal protein contacts with distance measurements.
Profiles PURE-system translation kinetics per codon (dwell times, elongation rates) grounded in published in-vitro reconstitution data.
Analyzes disome-seq ribosome collision risk per mRNA, flagging P-site/A-site codon pairs at elevated collision risk from published disome-sequencing occupancy data.
Models ribosome assembly thermodynamics with a stage-ordered maturation pathway, per-step ΔG and factor/r-protein binding measurements from published biochemical studies.
Resolves any free-form tissue to a PMID-grounded delivery route and LNP composition via the full-scale tissue resolver.
Runs deterministic drug repurposing queries that trace disease mechanisms to vulnerable codons.
Verifies coding-sequence structural soundness in the gene-editing and synthetic-biology designers: a clean ORF receives a green confirmation; a partial or non-canonical input (missing ATG start, no terminal stop, ragged length) receives an amber advisory with per-violation detail so users know the result was computed on a non-canonical sequence.
Simulates multi-agent treatment regimens where the agent picker spans PubChem compounds, approved drugs, and all designer-produced therapies (siRNA, ASO, mRNA, CRISPR, base-edit, prime-edit, ADAR guide, LNP formulations) materialised in the temporal graph — the simulator resolves mechanism-level effects for every agent type.
Designs gene edits (CRISPR sgRNA, base editor, prime editor) with ClinVar integration.
Designs synthetic biology constructs and codon-optimized gene therapy sequences.
Traces mechanism causation from gene to phenotype across the biological hierarchy.
Analyzes RNA modification combinatorial codes (the epitranscriptome).
Provides safety, toxicity, dosing, monitoring, and treatment-sequencing protocols.
Produces a coverage-confidence score for every disease so users see exactly where evidence is thick or thin.

Data foundations — 22 published databases

MODOMICS, RMBase, NCBI RefSeq, Ensembl, Ribo-seq, ClinVar, COSMIC, GEO, GTEx, GWAS Catalog, ClinicalTrials.gov, HPO (Human Phenotype Ontology), AlphaFold, UniProt, STRING DB, KEGG, Reactome, IMPC, ChEMBL, PubChem, DGIdb, Open Targets. No simulated or synthetic data is used.

Main sections

Home — landing page and platform overview.
Section 1 — Network Atlas: 42 entity types across 6 biological system levels with toggleable connection map and Graph Explorer.
Section 2 — Validated Discoveries: codon-wobble proofs, ribosome heterogeneity, rRNA modification site registry with ViennaRNA secondary-structure folds and PDB 4UG0 3D distance annotations.
Section 3 — Biological Analysis: metabolic profiler, comorbidity calculator, codon optimizer, patient confounders.
Section 4 — Therapeutic Design: RNA therapeutic designer, ribosome therapeutic designer (five wired ribosome data layers), drug repurposing, drug interactions, drug chemistry, synthetic biology, gene editing, treatment sequencing; Cure Blueprint with structural-evidence transparency and reviewer audit exports.
Section 5 — Reference & Coverage: disease coverage confidence, monitoring protocols, reference materials.
Treatment Simulation: simulate compound effects on disease mechanisms.
Temporal Graph: interactive perspective-driven traversal of the biological knowledge graph.
About: project background, methodology, contact.

Citations and provenance

Every recommendation surfaced by the platform carries an evidence chain back to one or more of the 22 source databases. PubMed IDs (PMIDs) are required for inclusion in the rRNA modification site registry; entries lacking a resolvable PMID are dropped at parse time. A machine-readable briefing for AI assistants is published at /llms.txt; the full route map is at /sitemap.xml.