40+ mechanistic models15+ scientific toolsCurated harmonized data

Biologics R&D, connected. Data, design, analysis, and pharmacology.

Bring public data, computational tools, molecule design, experimental data analysis, and QSP and PK/PD modeling together in one platform.

No signup required to explore the model library.

https://mcp.unibiointelligence.com/mcp

Ubi built · exploratory QSP

Target-gated cargo QSP model

A mechanistic model developed by UniBio Intelligence from public target-engagement biology to explore high local activity with limited systemic engagement.

40+

mechanistic models

15+

scientific tools

21

connected services

50+

curated tables available

1,300+

therapeutic antibodies

5

species in single-cell data

Featured data resource

Explore the therapeutic antibody landscape.

Search 1,300+ therapeutic antibodies by target, indication, clinical development, and regulatory status—then inspect available sequences, structures, and released PK/PD or QSP models.

  • Targets and indications
  • Clinical and regulatory evidence
  • Sequences and structures
  • PK/PD and QSP models
Open Antibody Explorer (opens in a new tab)

Ubi Biologics

Models and tools for working scientists.

Ubi Biologics is a platform for mechanistic modeling, scientific computation, data analysis, and publication-quality figures.

Modeling

Explore the mechanistic model library.

The library includes implementations of published models and mechanistic systems developed by UniBio Intelligence from public biology and pharmacology. We also work with teams to build program-specific PK/PD and QSP models through consulting engagements. Inspect equations, states, parameters, scenarios, technical diagrams, and scope limits; then compare outputs and export figures without installing a modeling stack.

PK/PDPBPKTMDDQSPPopPKMBMAMIPD
mAbADCbispecific / TCEfragmentdegradergene-delivered
Browse 40+ models
01
Scientific basis
Literature, declared assumptions, model purpose, and supported interpretation.
02
Technical topology
Compartments, states, interactions, dosing routes, and outputs.
03
Curated scenarios
Reference and program-relevant conditions with bounded changes you can simulate.
04
Browser simulator
Run, compare, visualize, and export results from the model page.

A connected scientific workflow

Question, computation, simulation, figure.

1

Ask the target question

Use UbiMCP to query expression, genetics, dependencies, trials, and therapeutic antibodies together.

2

Run the scientific tool

Number, humanize, design, predict structure, or assess the candidate with UbiTools.

3

Simulate the biology

Choose a matching PK/PD or QSP model and compare exposure–response scenarios.

4

Review and export the result

Generate tables and publication-quality figures for review, reporting, and the next experiment.

UbiMCP

Your LLM knows biology. Now give it the data and tools.

UbiMCP is the access layer for the whole platform. Add one hosted MCP endpoint to Claude, Codex, Cursor, or another compatible client, then search biomedical data, run UbiTools, and reach Ubi Biologics modeling from the conversation you already use.

https://mcp.unibiointelligence.com/mcp
Ubi Bio
19 public biomedical services for expression, genetics, targets, trials, structures, labels, pathways, and more.
Ubi Data
Processed, normalized, and harmonized datasets stored in analysis-ready tables for direct, SQL-backed computation. The Therapeutic Antibody Explorer connects targets, indications, clinical development, regulatory status, sequences, structures, and available PK/PD or QSP models.
Open the Therapeutic Antibody Explorer
Ubi Biologics
UbiTools, modeling, project data, analysis, and scientific workflows through the same platform endpoint.

Worked example

One question, several datasets, a few minutes.

“Which cell types express this target in the tissue, is there a genetic association, is it a dependency in relevant lines, and has anything been tried clinically?”
  • Slice single-cell expression by tissue, disease, cell type, platform, or dataset.
  • Pull trial metadata without writing SQL or joining AACT tables by hand.
  • Cross-reference target, compound, and trial records in one plain-language workflow.
  • Return structured values and citations from the queried resources.
Claude using UbiMCP to query colorectal cancer single-cell datasets and cross-reference highly expressed genes

Ask one cross-dataset question

Claude finds the relevant single-cell datasets, filters malignant epithelial cells, and calls UbiMCP resources in plain language.

Connected biomedical resources

  • Antibody Registry
  • ChEMBL
  • ClinicalTrials.gov
  • DailyMed
  • Ensembl
  • GTEx
  • MyChem.info
  • MyDisease.info
  • MyGene.info
  • NCBI
  • OmniPath
  • openFDA
  • Open Targets
  • Orange Book
  • PDB
  • Human Protein Atlas
  • Reactome
  • SureChEMBL
  • UniProt

Generated from the UbiMCP server registry. Ubi Data and Ubi Biologics capabilities sit alongside these public biomedical services.

UbiMCP + Claude Science

Enhance Claude Science with biologics-specific data, tools, and modeling.

Claude Science can work with the web, connected services, and multiple MCP servers. UbiMCP is designed to coexist with those connections, adding one endpoint for normalized biomedical data, 15+ scientific tools, and 40+ mechanistic models within the same scientific workflow.

Web, APIs, and public MCP servers
Complementary routes for discovery and record-level access. Dataset-scale questions can still require repeated requests and client-side joins.
Ubi Data
Normalizes and harmonizes biomedical sources into analysis-ready tables that support direct SQL-backed computation across full datasets.
UbiMCP
Adds Ubi Data, UbiTools, and mechanistic modeling to Claude Science through one biologics-focused endpoint that can sit alongside other MCP servers.
GTEx example: identifying tissue-specific genes through a narrow public API can require thousands of requests plus client-side aggregation. Ubi Data keeps the normalized expression tables analysis-ready, so the full dataset can be queried directly and joined with other biomedical sources in the same live analysis.

Same CD98hc question in Claude, with and without UbiMCP

Comparison of a structured UbiMCP answer with a web-search answer that incorrectly lists placenta as a GTEx tissue
With UbiMCP enabled, Claude queries the connected GTEx data. The web-only answer subtly conflated GTEx with another tissue resource.

Claude corrects the claim when challenged

LLM correction acknowledging that GTEx does not sample placenta
The initial statement sounded plausible, but the dataset boundary was wrong. Direct access to the normalized data makes that boundary explicit from the start.

From UniBio Intelligence

News and research

Browse all articles

Company news · Research grant

UniBio Intelligence Selected for Anthropic’s Rare Disease Research Program

Anthropic’s AI for Science program will support RareDose, our project connecting source-traceable evidence with pharmacology modeling for rare disease research.

Read the announcement
CUREBench bronze medal at NeurIPS 2025
NeurIPS 2025
CUREBench evaluates medical question answering, clinical reasoning, and the use of external biomedical tools.
Research
How we assessed 70,697 supplied protein models in CASP17, why the origin of a prediction matters, and what changing a shortlist can teach us before an experiment.
Research
How we approached both OpenADMET CYP inhibition tasks, and what unsuccessful experiments in molecular learning, docking, and data curation taught us along the way.

Expert services

We work with your team when the problem needs more than software.

Bring us a program-specific modeling, data-integration, or computational biology problem. We can adapt platform workflows, build custom analyses, and work alongside your scientists.

Meet with our CTO (opens in a new tab)

Prefer to explore first? Open Ubi Biologics.