Explority MCP

Explority MCP for querying early-stage orphan therapies described in the literature, ranked by success potential (predicted by explority.ai), using natural language queries such as:

• "Show gene therapies in oncology with the highest success potential"
• "Give me early-stage drugs for hypophosphatasia or Pompe disease"
• "Emerging therapies targeting VEGFR3"

Table of Contents

• Use Cases
• Overview
• Quick Start
• Query Examples
• Output Format
• How It Works
• Datasets Coverage

Use Cases

Explority MCP gives AI agents the ability to source emerging therapies with the highest success potential for:

• Rare disease drug discovery research
• Early-stage biotech scouting
• Competitive landscape analysis
• Target identification
• Therapy pipeline benchmarking

Overview

This server processes natural language queries about orphan therapies and returns filtered results from two datasets covering 5,846 rare diseases:

• First-in-class therapies (~15,000 therapies)
• Next-in-class therapies (~57,000 therapies)

All therapies are ranked by success potential based on evidence from scientific literature.

The dataset is updated monthly, with approximately 50,000 new articles processed each month.

An overview of diseases, along with the count of related therapies and therapies outperforming the average success rate, can be found in data/diseases.csv.

Quick Start

The server is available over HTTP with bearer token authentication.

1. Connect to hosted server

Claude Code

claude mcp add orphan-therapy --transport http https://app.explority.ai/mcp \
  --header "Authorization: Bearer explority_..."

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "orphan-therapy": {
      "transport": "http",
      "url": "https://app.explority.ai/mcp",
      "headers": {
        "Authorization": "Bearer explority_..."
      }
    }
  }
}

Codex / other MCP clients

{
  "mcpServers": {
    "orphan-therapy": {
      "type": "url",
      "url": "https://app.explority.ai/mcp",
      "headers": {
        "Authorization": "Bearer explority_..."
      }
    }
  }
}

2. Send queries

Then use the query_orphan_therapies tool with natural language queries.

Query Examples

* Show gene therapies in oncology with the highest success potential
* Give me early-stage drugs for rhabdoid tumor or Pompe disease
* Emerging therapies targeting VEGFR3
* Cell therapies for rare genetic diseases

Output Format

Returns a JSON-formatted response containing both first-in-class and next-in-class therapies from a single query:

{
  "first_in_class": {
    "name": "Top-10 emerging first-in-class orphan therapies",
    "therapies": [
      {
        "disease": "Disease name",
        "therapy": "Therapy type",
        "percentile": 2.77,
        "links": [
          "https://doi.org/10.1016/j.example.2025.01.001",
          "https://doi.org/10.1007/s12345-025-00123-x"
        ]
      }
    ]
  },
  "next_in_class": {
    "name": "Top-10 emerging next-in-class orphan therapies",
    "therapies": [
      {
        "disease": "Disease name",
        "therapy": "Therapy type (agent/target/mechanism)",
        "percentile": 15.49,
        "links": ["https://doi.org/10.1016/j.example.2025.02.002"]
      }
    ]
  }
}

If a category has no results, the therapies array will be empty. DOI strings are automatically parsed and converted into full URLs.

How It Works

Query processing flow:

1. MCP server extracts filters from a natural language query (disease types, disease names, therapy types, targets)
2. Applies filters to the explority.ai databases
3. Selects and reranks the top 10 relevant first-in-class and next-in-class therapies with the highest success potential
4. Returns formatted results with success percentiles

How probability of success is predicted:

1. A State-of-the-art algorithm structures published research and links scientific evidence to downstream clinical outcomes.
2. Large language models (LLMs) are trained as outcome-based classifiers on this structured dataset to predict the probability of success for novel first-in-class and next-in-class therapies.

Training data includes:

• 1M+ drug discovery articles
• connected to 10,708 orphan designations and 1,288 approvals

This enables Explority AI to identify evidence patterns associated with successful therapies and predict the success potential of new early-stage candidates - read the full article.

Datasets Coverage

| Dataset | Records | | ------------------------------------------- | -------: | | First-in-class therapies | ~15,000 | | Next-in-class therapies | ~57,000 | | Therapies with above avg. success potential | top 8.1% | | Rare diseases covered | 5,846 | | Articles processed monthly | ~50,000 | | Update frequency | Monthly |