Wheelder Cancer Research

1. What Is Wheelder Cancer Research?

Wheelder Cancer Research is a specialized AI platform built exclusively for cancer research and clinical inquiry. Unlike general-purpose AI assistants (ChatGPT, Claude, Gemini), this platform is deeply trained on cancer-specific biomedical data sources and implements Wheelder's proprietary Deep Research methodology.

The platform integrates 30+ peer-reviewed biomedical APIs and databases — from PubMed's 35M+ literature abstracts to ClinicalTrials.gov's 400K+ active trials, UniProt's protein data, Reactome's pathway information, and specialized cancer genomics databases like cBioPortal and GDC/TCGA.

When a researcher asks a question, Wheelder Cancer Research:

1. Classifies the query to understand the research intent
2. Fetches live data from 1-3 relevant biomedical APIs
3. Augments the AI prompt with real research data
4. Generates an evidence-based answer with citations to primary sources
5. Offers Deep Research workflows for multi-stage hypothesis testing
6. Recommends specialized AI tools (Claude, ChatGPT, Gemini, Grok, etc.) for different research phases

This ensures researchers get answers backed by actual research, not AI hallucinations.

2. Why Cancer-Specific?

Cancer research is one of the most critical and complex domains in biomedicine. General-purpose AI assistants lack the specialized training, real-time data access, and research methodology needed for cancer research. Wheelder Cancer Research addresses five critical gaps:

3. Deep Research Methodology

Wheelder invented the Deep Research methodology — a systematic approach to multi-stage research inquiry that goes beyond single-turn Q&A. Deep Research enables researchers to:

• Ask complex, multi-part questions and receive structured answers
• Refine hypotheses through iterative questioning
• Access specialized tools for different research phases (literature review, data analysis, visualization, collaboration)
• Build research workflows that combine multiple AI tools and data sources
• Preserve context across conversation sessions

In Wheelder Cancer Research, the "Deep Research" button triggers a workflow that:

1. Analyzes the current research question and answer
2. Generates a multi-stage research workflow
3. Recommends specialized AI tools (Claude for reasoning, ChatGPT for breadth, Gemini for multimodal analysis, etc.)
4. Opens a modal with tool recommendations and direct links
5. Allows researchers to continue their work in specialized platforms

4. 30+ Biomedical Data Sources

Genomics & Protein Data (8 sources)

• UniProt — Protein sequences, functions, and interactions (500K+ proteins)
• Reactome — Biological pathway database (2,500+ pathways)
• RCSB PDB — 3D protein structures (200K+ structures)
• Ensembl — Genome browser and variation data
• NCBI Gene — Gene information and sequences (30K+ human genes)
• InterPro — Protein families and domains (40K+ families)
• STRING — Protein-protein interaction networks (24M+ interactions)
• AlphaFold — AI-predicted protein structures (200M+ structures)

Clinical & Research Data (6 sources)

• PubMed — 35M+ biomedical literature abstracts
• EuropePMC — European biomedical literature (50M+ articles)
• ClinicalTrials.gov — 400K+ active clinical trials
• GWAS Catalog — Genome-wide association studies (200K+ associations)
• ClinVar — Genetic variant interpretations (1M+ variants)
• dbSNP — Single nucleotide polymorphisms (600M+ SNPs)

Drug & Medication Data (5 sources)

• OpenFDA — FDA drug approvals and adverse events
• DailyMed — Medication labels and prescribing information
• RxNorm — Drug naming and relationships (20K+ drugs)
• ChEBI — Chemical entities and structures (200K+ chemicals)
• DrugInteractions — RxNav drug interaction database

Disease & Health Data (11 sources)

• OpenTargets — Drug targets for diseases (20K+ targets)
• GDC/TCGA — Cancer genomics data (2.5 petabytes)
• cBioPortal — Cancer genomics portal (300+ studies)
• Orphanet — Rare disease information (6,000+ diseases)
• MedlinePlus — Consumer health information
• WHO GHO — Global health observatory data
• Human Protein Atlas — Protein expression in tissues (20K+ proteins)
• WikiPathways — Community-curated pathways (800+ pathways)
• LitCovid — COVID-19 literature tracking
• Reactome — Pathway analysis (already listed above)
• Additional specialized cancer databases

5. Feature Breakdown

6. Architecture & Technology Stack

User Query (Cancer Research Question)
    ↓
Query Classifier (analyzes intent, identifies research type)
    ↓
DataSourceManager (selects 1-3 relevant biomedical APIs)
    ↓
Parallel API Fetches (cached 24h in SQLite)
    ├─ PubMed, ClinicalTrials, UniProt, Reactome, etc.
    └─ Non-fatal: if API fails, Groq still answers
    ↓
Augmented Prompt (original query + real API data)
    ↓
Groq LLaMA 3.3 70B (generates evidence-based answer)
    ↓
Answer Summarization (1-2 sentence summary)
    ↓
Image Generation (Pollinations AI from summary)
    ↓
Deep Research Workflow (optional multi-stage refinement)
    ├─ Generate workflow steps
    ├─ Recommend specialized AI tools
    └─ Open modal with tool links
    ↓
User sees: Answer + Sources + Recommended Tools + Generated Image

7. API-Augmented Generation (AAG)

API-Augmented Generation is Wheelder's proprietary methodology that combines real biomedical data with AI generation. Instead of relying solely on training data (which becomes stale), AAG:

1. Fetches live data from authoritative biomedical APIs (PubMed, ClinicalTrials, UniProt, etc.)
2. Injects data into the prompt as context for the AI model
3. Generates answers grounded in real, current research
4. Cites sources so researchers can verify and explore further

This approach eliminates hallucinations and ensures answers are always backed by evidence.

8. Scientific Foundations

Wheelder Cancer Research is built on peer-reviewed research in AI, information retrieval, and cancer biology:

9. Comparison with Existing Platforms

10. Public Repository & Collaboration

Wheelder Cancer Research is open source and publicly available. We believe cancer research should be a collaborative effort. We welcome contributions, feedback, and partnerships with academic institutions, clinical researchers, and other organizations working on cancer research.

Our vision is to work together with Google's cancer research initiatives, OpenAI's health tools, academic institutions, and clinical researchers to unify our contributions under one ceiling and make a meaningful difference in accelerating cancer research and saving lives.