Data Model

How all 28 sources connect through the mapping tables.

Network Diagram

graph LR
    %% Hub nodes
    PM[("🧬 protein_map<br/>26,499 proteins")]:::hub
    DM[("🏥 disease_map<br/>31,884 diseases")]:::hub
    DrM[("💊 drug_map<br/>42,939 drugs")]:::hub

    %% Protein sources (direct UniProt join)
    UP[UniProt]:::direct --> PM
    RT[Reactome]:::direct --> PM
    CH[ChEMBL]:::direct --> PM
    IP[InterPro]:::direct --> PM
    CP[Complex Portal]:::direct --> PM
    SL[SomaLogic]:::direct --> PM
    MB[MarkerDB]:::direct --> PM
    HPA[HPA/Olink]:::direct --> PM
    PDB[PDB/SIFTS]:::direct --> PM

    %% Protein sources (Ensembl bridge)
    OT[Open Targets]:::bridge --> PM
    ST[STRING]:::bridge --> PM
    GT[GTEx]:::bridge --> PM
    DIS[DISEASES 2.0]:::bridge --> PM
    EN[Ensembl]:::bridge --> PM
    GN[gnomAD]:::bridge --> PM

    %% Disease sources
    OT --> DM
    DIS --> DM
    CV[ClinVar]:::direct --> DM
    CT[ClinicalTrials]:::direct --> DM
    FDA[OpenFDA]:::direct --> DM
    ICD[ICD-10-CM]:::direct --> DM
    UKB[UK Biobank]:::direct --> DM
    MH[MeSH]:::direct --> DM

    %% Ontology hierarchy feeds disease_map
    MO[MONDO]:::onto --> DM
    DO[Disease Ontology]:::onto --> DM
    EF[EFO]:::onto --> DM
    GO[Gene Ontology]:::onto --> PM

    %% Drug sources
    CH --> DrM
    TTD[TTD]:::direct --> DrM
    PC[PubChem]:::direct --> DrM
    FDA --> DrM

    %% Competitors (fuzzy match via gene name)
    MSD[MSD]:::comp -.-> PM
    QX[Quanterix]:::comp -.-> PM
    AL[Alamar]:::comp -.-> PM
    NM[Nomic]:::comp -.-> PM

    %% LLM-extracted KG
    KG[("🤖 LLM-extracted<br/>Knowledge Graph")]:::kg --> PM
    KG --> DM

    %% Cross-links between hubs
    PM <--> DM
    PM <--> DrM
    DM <--> DrM

    %% Styles
    classDef hub fill:#1a73e8,stroke:#fff,color:#fff,font-weight:bold
    classDef direct fill:#34a853,stroke:#333,color:#fff
    classDef bridge fill:#fbbc04,stroke:#333,color:#000
    classDef onto fill:#9c27b0,stroke:#333,color:#fff
    classDef comp fill:#ff6d00,stroke:#333,color:#fff
    classDef kg fill:#e91e63,stroke:#333,color:#fff

Legend

Color	Meaning	Join type
🟢 Green	Direct UniProt/MONDO ID in the data	Exact ID match
🟡 Yellow	Uses Ensembl IDs (bridged via protein_map)	ID mapping lookup
🟣 Purple	Ontology (provides the hierarchy + xrefs)	Parsed from OBO
🟠 Orange	Competitor data (protein names, not IDs)	Fuzzy name match
🔴 Pink	LLM-extracted from papers	Entity resolution
🔵 Blue	Mapping hub table	Central join point

Join Paths

Protein → Disease (6 sources)

graph LR
    P[Protein<br/>UniProt ID] --> PM[protein_map]
    PM -->|ensembl_gene_id| OT[Open Targets]
    PM -->|ensembl_protein_id| DIS[DISEASES 2.0]
    PM -->|gene_name| CV[ClinVar]
    PM -->|uniprot_id| MB[MarkerDB]
    OT -->|efo_id| DM[disease_map]
    DIS -->|doid| DM
    CV -->|medgen_id| DM
    MB -->|disease_name| DM
    DM --> D[Disease<br/>MONDO ID]

Protein → Drug (4 sources)

graph LR
    P[Protein] --> PM[protein_map]
    PM -->|uniprot_id| CH[ChEMBL]
    PM -->|uniprot_id| TTD[TTD]
    PM -->|entrez_gene_id| PC[PubChem]
    CH --> DrM[drug_map]
    TTD --> DrM
    PC --> DrM
    DrM --> D[Drug]

Protein → Protein (PPI, 3 sources)

graph LR
    P1[Protein A] --> PM[protein_map]
    PM -->|string_id| ST[STRING<br/>score ≥ 700]
    PM -->|uniprot_id| CP[Complex Portal<br/>same complex]
    PM -->|uniprot_id| KG[LLM KG<br/>INTERACTS_WITH]
    ST --> PM2[protein_map]
    CP --> PM2
    KG --> PM2
    PM2 --> P2[Protein B]

Identifier Flow

flowchart TD
    subgraph "Raw Data (28 sources)"
        A[UniProt IDs] 
        B[Ensembl IDs]
        C[Gene Names]
        D[MONDO/EFO/DOID]
        E[MeSH/ICD-10]
        F[ChEMBL/PubChem]
    end

    subgraph "Mapping Layer"
        PM[protein_map<br/>9 columns]
        DM[disease_map<br/>10 columns]
        DrM[drug_map<br/>5 columns]
    end

    subgraph "Unified Query"
        V1[unified_protein_disease]
        V2[unified_ppi]
        V3[unified_drug_target]
        V4[unified_expression]
    end

    subgraph "Knowledge Graph"
        N[Neptune<br/>201K mapping edges<br/>+ LLM extractions]
    end

    A --> PM
    B --> PM
    C --> PM
    D --> DM
    E --> DM
    F --> DrM

    PM --> V1
    PM --> V2
    PM --> V3
    PM --> V4
    DM --> V1
    DrM --> V3

    PM --> N
    DM --> N
    DrM --> N