--- title: "AI x Antibodies Hackathon Oslo: the evidence base for your build" description: "Amass is the evidence base at Nucleate Norway's AI x Antibodies Hackathon at Oslo Cancer Cluster, 14 November 2026. $100 in API credits per team and an Amass track." url: https://amass.tech/oslo-ai-antibodies-hackathon --- # AI x Antibodies Hackathon Oslo: the evidence base for your build AI x Antibodies Hackathon, Oslo Cancer Cluster, 14 November 2026 # Build the tool a drug discovery team would actually trust. Amass is the evidence base at the [AI x Antibodies Hackathon](https://oslo-ai-antibodies-hackathon.vercel.app/), a one-day build in Oslo run by Nucleate Norway. Every team gets **$100 in Amass API credits**, and there is an Amass track where the most creative use of the data wins extra credits. Amass is one REST API over six cross-linked \*Cores: 43M+ papers, 1.2M+ clinical trials, 22K+ molecules, 43K+ genes, FDA and EMA authorizations, and patents. Every answer arrives with the record it came from. [Create your account →](https://platform.amass.tech)[See the antibody data ↓](#data) Saturday 14 November, 09:00 to 20:00 at Oslo Cancer Cluster Innovation Park, Ullernchausséen 64/66, 0379 Oslo. Doors open at 08:30 with breakfast. Rules and logistics are on the [event page](https://oslo-ai-antibodies-hackathon.vercel.app/). Amass is joining remotely. Questions for us: [find us on Discord](https://discord.gg/sEGaBHMhWa). Quickstart ## Hand your agent the Amass skill Paste one prompt into Claude Code, Cursor, Codex, or any agent that reads skills. It installs the amass-api skill, which teaches the agent the six \*Cores, their filters, and the response envelope before it writes a single call. The skill reads your key from `AMASS_API_KEY`. Agent prompt [Agent setup](https://platform.amass.tech/documentation/for-ai-agents/llm-quick-reference) Install the Amass API skill. Run \`npx skills add amass-technologies/public-skills --skill amass-api\` and select the agent you are running in. You can read the skill directly at https://github.com/amass-technologies/public-skills/blob/main/skills/amass-api/SKILL.md (raw: https://raw.githubusercontent.com/amass-technologies/public-skills/main/skills/amass-api/SKILL.md), with the full API reference next to it in AMASS.md. The skill reads my key from AMASS\_API\_KEY. If it is not set, tell me to create a key at https://platform.amass.tech and export it. Then use the Amass API skill when working on this project. [SKILL.md](https://github.com/amass-technologies/public-skills/blob/main/skills/amass-api/SKILL.md)[View on GitHub](https://github.com/amass-technologies/public-skills/tree/main/skills/amass-api) The antibody data ## What is already known, before you design anything Running a protein model is the easy part. The hard part is knowing what a drug discovery team already believes about your target and about your method. These are real records, as the API returns them. DrugCore · a target with clinical precedent ### Five drugs that act on FcRn (FCGRT) FcRn recycles IgG and extends its half-life. Blocking it clears pathogenic antibodies, which is why the first approvals came in myasthenia gravis. - Efgartigimod alfaProteinApproval Vyvgart. 4 approved and 26 investigational indications. - RozanolixizumabAntibodyApproval Rystiggo. Approved for myasthenia gravis. - NipocalimabAntibodyApproval Approved for myasthenia gravis, 11 investigational indications. - BatoclimabAntibodyPhase 3 10 investigational indications. - OrilanolimabAntibodyPhase 2 6 investigational indications. BioMedCore · the methods you are up against ### Developability prediction, read closely Four papers worth reading in your first hour. The last one is the reason to be sceptical of any single score. - Five computational developability guidelines for therapeutic antibody profilingPNAS 2019PMID 30765520 Introduces the Therapeutic Antibody Profiler, a common baseline. - DeepSP: Deep learning-based spatial properties to predict monoclonal antibody stabilityComput Struct Biotechnol J 2024PMID 38827232 Sequence-only surrogate for MD-derived aggregation and charge scores. - PROPERMAB: in silico prediction of antibody developability using machine learningmAbs 2025PMID 40042626 Predicts HIC retention time and high-concentration viscosity. - Fitness Landscape for Antibodies 2: Protein AI models cannot yet consistently predict developability propertiesbioRxiv 2025PMID 41497662 30 models across 32 studies. Most do not correlate with most datasets. The call behind the picture bash ``` curl "https://api.amass.tech/api/v1/cores/genecore/records\ ?query=FCGRT&include=referencesDrugCore&limit=1" \ -H "Authorization: Bearer amass_YOUR_KEY" ``` response · 200 OK · GeneCore FCGRT ``` symbol FCGRT ENSG00000104870 19q13.33 365 aa function binds IgG and recycles it, extending its half-life in circulation tractability antibody: Approved Drug (clinical) constraint LOEUF 0.579 pLI 0.769 (LOEUF decile 2) essentiality not essential, 0 dependent lines of 1,258 screened structure 3D: yes · PDB 1EXU, 3M17, 4N0U and 27 more drugs 5 linked in DrugCore ``` Where to start ## Eight example use cases, and a prompt to paste The challenge has four directions, and you can pick one or mix them. Each idea below sits on records the \*Cores return today, so you can check the data layer in your first hour. Improve a measurable property of an antibody ### Half-life by design FcRn recycles IgG and sets how long it stays in circulation. Pull the FCGRT structures from GeneCore and the engineering papers from BioMedCore, then score your variants against what has already been published. GeneCore and BioMedCore, half a day ### Viscosity and aggregation at high concentration Subcutaneous dosing needs high-concentration formulations that stay fluid. Read how published predictors define and measure viscosity, then build an optimizer that proposes edits and shows the trade-offs it made. BioMedCore, one evening Identify weaknesses or failure modes in existing computational tools ### Re-run a benchmark with germline distance controlled One recent benchmark reports that germline edit distance accounts for about 40% of the apparent predictive power of protein language models. Build the check that tells a user how much of a score is just similarity to a germline. BioMedCore, one evening ### Which tool papers have been replicated For each predictor, list the papers that cite it, whether any are retractions, and whether anyone reports it on a test set the authors did not choose. Publish the table with the gaps marked. BioMedCore, a few hours Help automate part of the antibody design workflow ### A target dossier agent Give it a gene symbol and get one page back: tractability and safety liabilities from GeneCore, the drugs against it from DrugCore, the trials that ran them from TrialCore, and the labels from RegulatoryCore. Every line carries its record id. all six Cores, half a day ### What regulators said about immunogenicity RegulatoryCore searches the parsed text of FDA labels, reviews and EMA assessment reports. Ask it what was said about anti-drug antibodies for approved antibodies similar to yours, and feed that into your liability check. RegulatoryCore, a few hours Explore new ways to use AI for computational drug discovery ### A clinical-stage reference set Developability guidelines were derived from the values seen in clinical-stage antibodies. Build your own reference set from DrugCore by modality and stage, with links to trials and labels, and score candidates against it. DrugCore, half a day ### A patent landscape for a target and format PatentCore is searchable over claims and carries CPC codes and family links. Map who has filed on a target and an antibody format, which filings are cited most, and where the white space is. A landscape, not legal advice. PatentCore, a few hours Paste this into Claude or Cursor to start prompt ``` Map the clinical precedent for an antibody target. Use the Amass API at api.amass.tech with my key in AMASS_API_KEY. Look up the gene FCGRT in GeneCore with include=referencesDrugCore, and read its antibody tractability. For each drug it links to, fetch the DrugCore record with include=referencesTrialCore and include=referencesRegulatoryCore. Give me one row per drug: modality, highest clinical stage, number of linked trials, and whether there is an FDA or EMA authorization. Then search BioMedCore for antibody developability prediction papers from 2022 onward, and tell me which two I should benchmark against. Say clearly where a link list is empty, because that means no link recorded, not no evidence. ``` The offer ## $100 in API credits per team, and an Amass track Create your account and API key on the platform before Saturday so the morning goes on building. [Create your account →](https://platform.amass.tech)[Open the build assistant](https://platform.amass.tech/assistant) What $100 buys Roughly **2,000 searches** or **10,000 record fetches**. Search costs $0.05 per 20 results; get-by-id and lookup are $0.01 each. That is plenty for one day, so query freely and cache what you re-run. The Amass track Use Amass in your project and tell the judges how. The most creative use of the data wins extra credits on top of your team’s $100. Show which records your tool read and what it did with them. Apps to fork ## You do not have to start from an empty file Every app in the gallery runs on the same API, the same key, and the same cited records as the examples above. All of them were built with [Lovable](https://amass.tech/life-science-api), so there is a route through this hackathon that involves no local setup at all. Fork one, point it at your key, and change the question it asks. [Browse the app gallery →](https://platform.amass.tech/app-gallery)[No-code recipes](https://amass.tech/life-science-api) platform.amass.tech/app-gallery Research #### From Gene to Approval One gene → every drug, trial, and FDA/EMA approval behind it. Research #### Graveyard & Garden 25 years of late-stage trials, halted or still alive. Research #### Clinical Trial Atlas Where a drug is tested, and which regions are enrolling. Research #### Indication Timeline A drug across its Phase-3 indications and approvals. Regulatory #### Transatlantic Divergence Desk FDA versus EMA outcomes, side by side. Clinical #### Pipeline Monitor A weekly digest of new Phase 2/3 trials by sponsor. Nineteen starter apps across research, regulatory, and clinical. Each one is a working answer to “what does a small app on this data look like?” Find us ## We are remote, and on Discord Amass is not at the venue. Ask us anything in the Amass Discord, even if the build is still a sketch. Discord Questions about credits, endpoints, or which \*Core holds what are all fair game. Post the query you tried and what came back, and we will help from there. [Join the Amass Discord →](https://discord.gg/sEGaBHMhWa) Before the day Create your account and API key on the platform, and try the prompt above, so Saturday morning goes on building. The build assistant can write your first call with you. Playbook ## Six things worth knowing before the kickoff These are what separate an entry that lands from one that spends the day fighting its own plumbing. 01 ### Narrow until it feels too small. You have roughly eight hours of building. A finished small tool beats an unfinished large one, and the judges can tell. 02 ### Read the prior art in the first hour. Developability and antibody design both have a deep literature. Twenty minutes with BioMedCore saves you from rediscovering a 2019 result at 16:00. 03 ### Build the baseline first. A random score, a simple biophysical rule, the manual version of the workflow. It turns your number into a result instead of a claim. 04 ### Show where it fails. A demo that names its own failure mode reads as principled work, and it is the part judges remember. 05 ### Cite it or drop it. Every Amass record carries a stable id and a route back to its source, so there are no invented PMIDs in your pitch. 06 ### Read the envelope, respect the limit. Results arrive under data and failures under error. The limit is 60 requests per 60 seconds; on a 429, read Retry-After and back off. There is no paging, so narrow with filters. Prototype in Claude ## Sketch it in a chat before you write any app code. Connect the Amass MCP in Claude, Cursor, ChatGPT, or Codex and query all six \*Cores in natural language. Look up what is already known about an antibody target, find the filters that work, and only then drop the same queries into your build. On the evening it doubles as the demo: the answers come back cited in the chat, with nothing to deploy. [Connect the MCP](https://amass.tech/mcp)[Or install the Amass skill](#quickstart) ## Build something a drug discovery team would trust Set up your key before Saturday, and let the API do the reading. Narrow it until it feels too small, build the baseline, and show where it breaks. [Create your Amass account →](https://platform.amass.tech)[Amass Discord](https://discord.gg/sEGaBHMhWa)[Connect the MCP](https://amass.tech/mcp) We are remote. Find us on the [Amass Discord](https://discord.gg/sEGaBHMhWa). Data No training on your data Hosting EU-hosted, enterprise Isolation Tenant-isolated by architecture Certification ISO 27001, SOC 2, NIS2: in motion