~1,400
Known Unique Folds
5
Most Common Superfolds
~10%
Enzymes are TIM Barrels
~20%
Proteome = Rossmann-like
TIM Barrel (α/β Barrel)
α/β Mixed
8 alternating α-helices and 8 parallel β-strands forming a barrel. The most common enzyme fold — ~10% of all enzymes use it. Found in 5 of 7 enzyme classes. The active site sits inside the barrel.
💊 Drug Discovery Gold: TIM barrels are ubiquitous in metabolic enzymes. Understanding their folding = designing inhibitors for cancer metabolism, infectious disease enzymes, and metabolic disorders. AI models predicting TIM barrel variants accelerate drug target ID.
RXRXSDGRRLAYMRKREGN
Rossmann Fold
α/β Mixed
Repeated β-α-β motifs forming a sheet flanked by helices. Binds nucleotide cofactors (NAD⁺, FAD, ATP). Makes up ~20% of all protein structures — the most common fold in nature.
🧬 Synthetic Bio Engine: Rossmann folds power oxidoreductases, kinases, and dehydrogenases. Engineering new Rossmann variants = custom enzymes for biomanufacturing, biofuels, and industrial chemistry. Key target for autonomous enzyme engineering.
TWSTDNACDXSCTVABAYN
Immunoglobulin Fold
β-Sheet
Two β-sheets packed face-to-face forming a "sandwich." The structural backbone of antibodies, T-cell receptors, and cell adhesion molecules. Critical for immune recognition.
💉 Antibody & Gene Therapy: Every therapeutic antibody is built on Ig folds. AI prediction of Ig fold variants = better antibody drugs, CAR-T therapies, and delivery vehicles for gene editing. Chai-2 designs de novo antibodies using this fold.
CRSPBEAMSANAREGNVRTXMRNA
Globin Fold
α-Helical
All-α bundle of 8 helices surrounding a heme group. Found in hemoglobin and myoglobin — the oxygen transport system. One of the first protein structures ever solved (1958).
🩸 Gene Therapy Target #1: Sickle cell disease is a single amino acid mutation in the globin fold. CASGEVY (CRSP/VRTX) cures it by editing the gene. Understanding globin folding dynamics = next-gen hemoglobin therapies.
CRSPVRTXBEAMNTLA
Helix-Turn-Helix (HTH)
α-Helical
Two α-helices connected by a short turn. The classic DNA-binding motif. Found in transcription factors, repressors, and gene regulators across all domains of life.
🧪 Gene Regulation & Editing: HTH motifs are how proteins find and bind specific DNA sequences. CRISPR-Cas proteins use related structural motifs. Predicting HTH binding specificity = designing better gene editors and transcription therapies.
CRSPNTLABEAMCRBUSGMO
α/β Plait (Ferredoxin-like)
α/β Mixed
A 4-stranded β-sheet with 2 α-helices packed on one side. One of the 5 most common superfolds. Found in ferredoxins, RNA-binding domains, and ubiquitin.
⚡ RNA Therapeutics: RNA-binding domains with this fold are targets for mRNA delivery optimization. Understanding plait fold dynamics = better mRNA drug stability (Moderna, BioNTech).
MRNAILMNNTRA
β-Propeller
β-Sheet
4-8 blade-shaped β-sheets arranged around a central axis like a propeller. Found in signaling proteins, receptors, and viral surface proteins. WD40 repeats are the most common.
🦠 Viral & Cancer Targets: Many viral entry proteins and cancer signaling hubs use propeller folds. Predicting propeller conformations = designing drugs against "undruggable" cancer targets and pandemic preparedness antivirals.
RXRXSDGRTEMAZNMRK
β-Trefoil
β-Sheet
12 β-strands forming a barrel capped by 6 hairpins in a trefoil pattern. Found in cytokines (IL-1, FGF) and toxins. Key role in inflammation and growth signaling.
🔥 Inflammation & Autoimmune: IL-1 family cytokines drive autoimmune diseases. Predicting trefoil fold variants = designing anti-inflammatory biologics and precision immunotherapy drugs.
REGNAZNVRTXRVMD
Coiled-Coil
α-Helical
Two or more α-helices wound around each other in a rope-like structure. Found in structural proteins (keratin, collagen), motor proteins (myosin), and viral fusion proteins.
🧱 Materials & Delivery: Coiled-coils are the basis for self-assembling nanostructures for drug delivery. Synthetic biology teams engineer coiled-coils as scaffolds. Also key to understanding viral fusion = vaccine design.
DNATWSTMRNASANA
Intrinsically Disordered Regions
Special / Disordered
~30-40% of eukaryotic proteins have disordered regions with NO stable fold. They adopt structure only when binding partners. Long considered "undruggable" — now a major frontier.
🎯 The "Undruggable" Frontier: Protein folding interference targets these transient states. p53 (mutated in >50% of cancers) has large disordered regions. New AI models that predict disorder dynamics = massive drug target expansion.
RXRXRLAYSDGRTEMGOOGL
GPCR 7-TM Bundle
α-Helical
7 transmembrane α-helices forming a bundle embedded in cell membranes. G protein-coupled receptors are the #1 drug target class — ~34% of all FDA-approved drugs target GPCRs.
💊 Biggest Drug Target Class: AlphaFold's ability to predict GPCR conformations is revolutionizing drug design. Understanding how these folds shift between active/inactive states = designing better drugs for pain, heart disease, CNS disorders.
GOOGLRXRXSDGRREGNMRKAZN
Amyloid Cross-β
β-Sheet
Misfolded β-sheets that stack into insoluble fibrils. The hallmark of Alzheimer's (Aβ), Parkinson's (α-synuclein), prion diseases, and ALS. Understanding misfolding = understanding neurodegeneration.
🧠 Neurodegeneration Goldmine: D-I-TASSER maps folding *pathways* — crucial for catching amyloid intermediates before fibrils form. This is the key to Alzheimer's and Parkinson's drugs. Trillion-dollar market if solved.
REGNMRKAZNRXRXGOOGL
How Fold Types Map to the Domino Chain
Each protein fold type connects to specific dominoes in the investment cascade. When AI solves a fold's dynamics, the corresponding domino advances — and the mapped stocks benefit.
D1 → D2: Structure Solved → Drug Target Identification
GPCR 7-TM Bundle→D2 Drug Target ID34% of all drugs target GPCRs. AlphaFold predicts conformations → faster hit-finding. GOOGL, RXRX, SDGR, MRK
Amyloid Cross-β→D2 Drug Target IDD-I-TASSER maps folding pathways → catches toxic intermediates in Alzheimer's, Parkinson's. REGN, AZN, RXRX
Disordered Regions→D2 Drug Target IDProtein folding interference unlocks "undruggable" targets (p53, MYC). RXRX, RLAY, TEM
β-Propeller→D2 Drug Target IDCancer signaling hubs & viral entry proteins → oncology + antivirals. SDGR, TEM, AZN
D3: Gene Editing & Delivery
Helix-Turn-Helix→D3 Gene EditingDNA-binding motifs in CRISPR-Cas. Predicting HTH specificity = better editors. CRSP, NTLA, BEAM, CRBU
Immunoglobulin Fold→D3 DeliveryIg folds are antibody scaffolds for targeted delivery. Chai-2 designs de novo antibodies. SANA, REGN, VRTX
Globin Fold→D3 Gene EditingSickle cell = globin mutation. CASGEVY already approved. Next: other hemoglobinopathies. CRSP, VRTX, BEAM
Coiled-Coil→D3 DeliverySelf-assembling nanostructures for LNP and viral vector delivery systems. SANA, MRNA, DNA
D4: Sequencing & Diagnostics
α/β Plait→D4 DiagnosticsRNA-binding domains → mRNA stability for liquid biopsy probes. ILMN, NTRA, GH
Amyloid Cross-β→D4 DiagnosticsDetecting misfolded protein biomarkers in blood → early Alzheimer's diagnosis. NTRA, GH, WGS
D5: AI Drug Discovery
TIM Barrel→D5 AI Drug Discovery10% of enzymes. AI virtual screening of TIM barrel inhibitors 1000x faster with Boltz-2. RXRX, SDGR, RLAY
GPCR 7-TM Bundle→D5 AI Drug DiscoveryAlphaFold3 predicts GPCR-ligand complexes → virtual screening for billions of compounds. GOOGL, RXRX, SDGR
β-Trefoil→D5 AI Drug DiscoveryCytokine targets for autoimmune → AI designs custom biologics against IL-1 family. REGN, AZN, RVMD
D6-D8: Synthetic Bio → Agriculture → Materials
Rossmann Fold→D6 Synthetic Bio20% of proteome. Engineering custom enzymes for biomanufacturing & biofuels. TWST, DNA, CDXS
TIM Barrel→D7 AgricultureCrop enzyme engineering — pest resistance, nitrogen fixation. Bayer uses OpenFold3. BAYN, CTVA
Coiled-Coil→D8 MaterialsSelf-assembling protein materials → spider silk, biodegradable plastics. DNA, TWST
AI Protein Folding Models — April 2026
The competitive landscape of AI models predicting protein structure, dynamics, and drug binding. Open-source vs proprietary is a key battle — OpenFold3 and Boltz-2 are democratizing access while AlphaFold3 remains restricted for commercial use.
AlphaFold 3
Google DeepMind (Isomorphic Labs)
Predicts protein + DNA/RNA + ligand complexes. Joint 3D structures of entire molecular complexes. Most accurate but commercially restricted.
Restricted Commercial Use
GOOGL
OpenFold3
OpenFold Consortium (Novo Nordisk, Bayer, Cyrus Bio, +)
Open-source AlphaFold3 alternative. Trained on 300K+ experimental structures + 13M synthetic. Commercially usable. Novo Nordisk, Bayer Crop Science, Outpace Bio already using it.
Open Source (MIT)
BAYNNVO
Boltz-2
MIT + Recursion Pharmaceuticals
First model combining structure + binding affinity prediction. 1000x faster than physics-based FEP. Ranked #1 on CASP16 affinity. Open-source MIT license.
Open Source (MIT)
RXRXNVDA
D-I-TASSER
University of Michigan
AI + physics hybrid. 13% better on complex proteins. Maps folding *pathways* not just final states — crucial for neurodegeneration targets and transient intermediates.
Academic Open
REGNMRK
Chai-2
Chai Discovery (ex-OpenAI)
De novo antibody design. Generates full-length antibodies with therapeutic attributes. Exclusive deal with Eli Lilly for biologics discovery.
Commercial (Lilly Exclusive)
LLY
Coefficient Bio Platform
Anthropic (acquired April 2026)
AI-driven drug R&D planning, clinical regulatory strategy, and candidate discovery. Acquired for $400M. Expected to become "Claude for Life Sciences."
Proprietary (Anthropic)
GOOGL
QuPepFold
Academic / Quantum Computing
Quantum computing approach to protein folding. Early-stage but accessible to structural biology teams. Could leapfrog classical AI models on large protein complexes.
Research Preview
GOOGLMSFTNVDA
AstraZeneca MapDiff
AstraZeneca + University of Sheffield
Proprietary inverse protein folding framework. Designs amino acid sequences for desired 3D structures. Big pharma building in-house AI capabilities.
Proprietary
AZN
Fresh Signals — April 2026
New breakthroughs, deals, and developments since the last intelligence scan. These signals feed directly into the domino chain and stock recommendations.
March 29, 2026
Eli Lilly × Insilico Medicine — $2.75B AI Drug Deal
Lilly pays $115M upfront + up to $2.75B in milestones for Insilico's AI-discovered drug candidates across oncology, metabolic disease, and immunology. Lilly gets exclusive license to develop, manufacture, and market. The largest AI drug discovery deal in history.
🔴 HIGH IMPACT — D5 AI Drug Discovery accelerating
LLYRXRXSDGRTEMRLAY
April 3, 2026
Anthropic Acquires Coefficient Bio for $400M
Frontier AI lab enters drug discovery directly. Coefficient Bio's founders came from Genentech/Prescient Design. Expected to become "Claude for Life Sciences" within 12-18 months. Signals that big AI companies see drug discovery as core market.
🔴 HIGH IMPACT — AI infrastructure domino (Big Tech entry)
GOOGLNVDAMSFT
January 2026
Pharma Platform Deal Wave: GSK + Noetik, Lilly + Chai, Pfizer + Boltz
Three major pharma companies signed platform-level AI deals in a single month. GSK pays $50M upfront to Noetik for cancer foundation models. Lilly integrates Chai-2 for de novo antibody design. Pfizer partners with Boltz for small molecule discovery. AI is now core R&D infrastructure, not an experiment.
🔴 HIGH IMPACT — D5 domino potentially upgrading to "ADVANCED"
RXRXSDGRRLAYTEM
Q1 2026
Boltz-2: Structure + Affinity in One Model (1000x Faster)
MIT + Recursion release Boltz-2 under MIT license. First model to jointly predict 3D structure AND binding affinity. Matches physics-based FEP accuracy but 1000x faster. Ranked #1 on CASP16 affinity. Open-sourced — anyone can use commercially.
🔴 HIGH IMPACT — Democratizes drug discovery computational power
RXRXNVDA
Q1 2026
OpenFold3: Open-Source AlphaFold3 Alternative Goes Commercial
Unlike AlphaFold3 (restricted), OpenFold3 is MIT-licensed for commercial use. Novo Nordisk, Bayer Crop Science, Outpace Bio, and Cyrus Biotechnology already deploying it. Federated training initiative with 5 pharma companies pooling proprietary data.
🟡 MEDIUM IMPACT — Broadens access but dilutes Google's moat
GOOGLBAYN
2026 Emerging
Protein Folding Interference — New Drug Mechanism Class
Targeting transient folding intermediates to induce protein degradation. Opens "undruggable" targets across oncology, neurology, and rare diseases. Acts on states that AlphaFold alone can't fully predict — needs pathway models like D-I-TASSER.
🟡 MEDIUM IMPACT — Early-stage but could be huge for undruggable targets
RXRXRLAYTEM
Ongoing 2026
RXRX at .886 Fibonacci — Deep Value Zone + Boltz-2 Catalyst
Recursion ($3.60) sits near the .886 retracement with analyst target $9.14. Boltz-2 open-source gives credibility. Sanofi/Roche/Bayer/Merck partnerships generating $500M+ in payments. But still burning cash with $74.68M revenue. High conviction, high risk.
🔴 WATCH — Deep value entry if clinical data validates
RXRX
2026 AI Drug Discovery Deal Tracker
Big pharma is spending billions to integrate AI into core R&D. These deals signal which companies and technologies are gaining institutional validation.
| Date | Deal | Value | Focus | Signal |
|---|---|---|---|---|
| Mar 2026 | Eli Lilly × Insilico Medicine | $2.75B | AI-discovered drug candidates (oncology, metabolic, immunology) | Largest AI drug deal ever. Lilly all-in on AI pipeline. |
| Apr 2026 | Anthropic × Coefficient Bio | $400M | AI drug R&D planning + clinical strategy | Frontier AI labs entering drug discovery directly. |
| Jan 2026 | GSK × Noetik | $50M upfront | Cancer foundation models (NSCLC, CRC) | Platform deal, not single-asset. AI as R&D infra. |
| Jan 2026 | Eli Lilly × Chai Discovery | Mid-8 figures/yr | De novo antibody design (Chai-2) | Infrastructure integration, not one-off. Custom AI model on Lilly data. |
| Jan 2026 | Pfizer × Boltz | Undisclosed | Small molecule drug discovery | Three big pharma deals in one month = inflection point. |
| Jan 2026 | Aurora Therapeutics Launch | $16M seed | Personalized gene editing (PKU first) | Doudna + Urnov + Menlo Ventures. Clinical trial 2027. |
| Early 2026 | Isomorphic Labs Phase 1 | — | First AI-designed cancer drug enters trials | Hassabis confirmed at Davos. Google's direct drug play. |
| 2025-26 | Recursion Partnerships | $500M+ cumulative | Sanofi, Roche, Bayer, Merck — AI platform access | Open-sourced Boltz-2 for credibility. Revenue growing but still burning cash. |
What This Means for the Domino Chain
The January 2026 deal wave — GSK/Noetik, Lilly/Chai, Pfizer/Boltz — marks an inflection point. Big pharma is no longer experimenting with AI; they're making it core R&D infrastructure. This accelerates D5 (AI Drug Discovery) significantly and validates the entire thesis that protein folding solutions create cascading investment opportunities.
The Anthropic/Coefficient Bio acquisition adds a new dimension: frontier AI companies are entering biotech directly. If "Claude for Life Sciences" materializes within 18 months, it could compress timelines across D2-D6 simultaneously. Watch for Google (Isomorphic Labs Phase 1 results) and Microsoft (potential biotech acquisitions) to follow.