Private beta · Now inviting design partners
Your personal research assistant for therapeutic discovery
Take a therapeutic hypothesis to cited evidence and ranked candidates, without an enterprise budget or a computational team.
- Every claim cited and traceable
- One connected biology and chemistry graph
- Plans from $39 a month, no procurement
Discovery tools are built for pharma. Most discovery starts smaller.
Bringing one drug to market costs an estimated $2.6 billion, yet the earliest and cheapest questions are the hardest for small teams to answer.
Priced for enterprises
Leading R&D platforms are sold through institutional contracts, procurement and security reviews.
Fragmented data
Literature, compounds, structures and assays sit in separate databases: PubMed, PubChem, ChEMBL, UniProt and PDB.
Costly outsourcing
Without an in-house computational team, early analysis means hiring consultants or contract research organizations.
Every program starts with the same four questions
Triangle BioTech gives you one research workspace, built on a connected knowledge graph of public biology and chemistry data, with an AI research agent that shows its work.
Is this target actionable?
Literature and target intelligence pulls the relevant evidence together in one place, instead of database by database.
What is already known about it?
The knowledge base connects genes, diseases, compounds, structures and papers in one graph.
Which compounds deserve a closer look?
The AI research agent ranks candidates and answers questions like "What evidence contradicts this?" with citations.
What experiment should come next?
Your research workspace keeps every hypothesis, result and decision, so your progress compounds.
Investigations KRAS G12D in pancreatic cancer
Start an investigation
Is KRAS G12D a tractable target in pancreatic cancer, and what should I test next?
KRAS G12D
Pancreatic ductal adenocarcinoma
Directly inhibiting KRAS G12D can control G12D-driven pancreatic tumors.
Searching literature, compound, structure and trial records
Target evidence
- Known
Up to 90% of pancreatic tumors carry an activating KRAS mutation.Peer-reviewed literature · PMC
- Known
G12D is the most common KRAS mutation, found in about 40% of patients.ASCO GI Symposium abstract, 2026
- Known
Daraxonrasib became the first approved RAS-targeted therapy for pancreatic cancer in 2026.FDA approval · RASolute 302, NEJM
- Inferred
RAS is a clinically validated target here; G12D-selective inhibition shows early clinical activity.From the findings above and Phase 1 trial data
Connected knowledge graph
Genes, diseases, compounds and signaling links from public databases, joined in one graph.
Candidates, ranked by clinical maturity Predicted
- DaraxonrasibMulti-selective RAS(ON) inhibitor
FDA approved, 2026OS 13.2 vs 6.7 months
- INCB161734G12D-selective, ON/OFF
Phase 137% response rate, n=41
- VS-7375G12D-selective, ON/OFF
Phase 1/2FDA Fast Track
- SetidegrasibG12D protein degrader
Phase 1Alone and with chemotherapy
Each row links to its trial record and publication.
Ask the research agent
What evidence argues against a G12D-selective inhibitor on its own?
Known In pancreatic cancer models, the G12D inhibitor MRTX1133 raised EGFR and HER2 signaling, a feedback loop that can blunt response.
Known Adding the pan-ERBB inhibitor afatinib was synergistic, and stayed active in MRTX1133-resistant cells.
Inferred Single-agent responses may be limited by adaptive receptor signaling.
Sources: Cancer Research, 2023 · J Gastrointest Oncol, 2024Plan the next experiment
Pairing G12D inhibition with pan-ERBB blockade will prevent signaling rebound in G12D pancreatic organoids.
- Arms
- MRTX1133 alone, afatinib alone, and the combination
- Inhibitor
- MRTX1133, a research-grade G12D inhibitor; clinical candidates aren't commercially available
- Models
- 3 KRAS G12D patient-derived organoid lines
- Readouts
- pERK at 24 and 72 hours, viability at day 5
Step 1 of 6: Start an investigation
Sample Investigation: Built from public sources, current as of September 2026. For research use only; not medical advice.
Lab Notebook ELN-0142 · G12D + pan-ERBB organoid assay
Write the protocol
Pairing G12D inhibition with pan-ERBB blockade will prevent signaling rebound in G12D pancreatic organoids.
- Day 0 Seed 3 KRAS G12D organoid lines in 96-well plates
- Day 1 Treat: MRTX1133 dose series ± afatinib, plus afatinib alone
- 24 h · 72 h Lysates from parallel 12-well cultures for pERK westerns
- Day 5 Read viability with CellTiter-Glo 3D, triplicate wells
Protocol saved as version 1 · status: Planned
Register materials
- MRTX1133KRAS G12D inhibitor · 10 mM in DMSOLot A-2291
- AfatinibPan-ERBB inhibitor · 10 mM in DMSOLot B-0874
- PDO-01, PDO-02, PDO-03KRAS G12D pancreatic organoidsPassage 8–11
- CellTiter-Glo 3DViability readoutExp. 03/2027
- Anti-phospho-ERK1/2Thr202/Tyr204 · with total ERK1/2 and GAPDH controls1:2000
Lay out the plate
- MRTX1133 alone
- + afatinib 100 nM
- Afatinib alone
- Vehicle
- Kill control
Plate 1 of 2 · rows A–C PDO-01, D–F PDO-02, in triplicate · PDO-03 on plate 2
Record the run
Seeded PDO-01 to 03 at 2,000 cells per well in Matrigel.
Dosing complete. Dispense log attached from the liquid handler.
24 h lysates collected. Organoid morphology normal in vehicle wells.
Observation Combination wells visibly smaller than single agent from Day 3.
Analyze the assay Simulated data
Review and sign
Plate quality passedZ′ = 0.71 across controls
Combination beat both single agents in all 3 linesIC50 shifted about 6-fold; afatinib alone had little effect
pERK rebound suppressed at 72 hWestern blot, all 3 lines, normalized to total ERK
Signed by author and witnessFull audit trail · 18 recorded changes
Step 1 of 6: Write the protocol
Sample Notebook: Illustrative workflow with simulated data. Not real experimental results.
Answers you can defend
AI is only useful in research if you can check it. Every result in Triangle BioTech is labeled by how much you should trust it, and linked to the source, model and version behind it.
- Known
Established in the published literature or curated databases, with citations.
- Inferred
Follows from connecting known facts across the knowledge graph.
- Predicted
Produced by a model, with the model and its version recorded.
- Hypothesis
Your own ideas and next steps, kept separate until you test them.
Start for $39 a month. No procurement required.
Self-serve plans for individual researchers, with compute-heavy work billed separately as credits.
Explorer
For students and early exploration
$39 / month
- 3 research projects
- Literature and target intelligence
- AI research assistant
- $5 in compute credits
Researcher
For serious independent researchers
$99 / month
- 10 research projects
- Full research graph and analytics
- Small virtual screens
- $25 in compute credits
Advanced
For founders and funded academic labs
$249 / month
- Unlimited projects
- Private datasets and API access
- Larger virtual screens
- $100 in compute credits
Working with a team? Research Lab plans start at $599 a month for five users, and Institutional plans are available for larger organizations. Tell us about your team.
Making serious discovery research accessible
We believe every scientist with a promising hypothesis should have the tools to test it, not just the teams with enterprise budgets. Triangle BioTech starts with the individual researcher and grows with them, from a single project to a lab to an institution.
Our platform is in early development today, and we are building it alongside our first design partners.
Christopher Schoppy
Founder and CEO
A technology leader who bridges business strategy and technical execution, with 30 years at the intersection of scientific computing, software development and enterprise cloud infrastructure.
Join the waitlist
We are opening the private beta to a small group of design partners first. Tell us about your research and we will be in touch.
Founding members get a discounted first year in exchange for early feedback that shapes the product.