Pathway Exploration
Load KEGG pathways, interact with entities, and resolve genes and compounds to 3D structures and data cards
Overview
KEGG pathway maps are interactive on the Omics-OS Cloud canvas. Load any pathway via the command bar, then explore it — single-click entities for AI awareness, double-click to resolve them into protein structures, compound cards, or linked pathways. This page walks through the complete pathway exploration workflow.
Loading a Pathway
- Open the command bar (press
/or click Search) - Select the Pathway tag (type "path" then press Tab)
- Search for a pathway (e.g., "glycolysis", "MAPK", "mTOR signaling")
- Select a result — an interactive PathwayNode (1200x800) appears on the canvas
Pathways are sourced from KEGG (Kyoto Encyclopedia of Genes and Genomes). Use standard KEGG pathway names for best results.
Interacting with Pathway Entities
Single-Click: AI Context
Single-clicking an entity (gene, enzyme, compound) in the pathway makes the AI assistant aware of it. Once clicked:
- The Lobster AI can reference the entity in subsequent conversations
- Useful for asking questions like "What does this gene do?" after clicking it
- The entity stays in the AI context until the conversation moves on
Double-Click: Entity Resolution
Double-clicking an entity triggers the entity resolution pipeline — the command bar opens pre-filled with the entity's information.
- Double-click a gene, enzyme, or compound in the pathway diagram
- The system determines the entity type (gene maps to the Protein tag, compound maps to the ChEMBL tag)
- The command bar opens pre-filled with the appropriate tag and search query
- Results appear immediately — select one
- A new canvas node is created (3D structure, compound card, or linked pathway)
Entity Type Resolution
How each entity type is resolved when you double-click it:
| Entity Type | Tag Used | Search Query | Canvas Result |
|---|---|---|---|
| Gene / Enzyme | Protein | Gene symbol or UniProt ID | MolstarNode (3D structure) or InfoCard |
| Compound (named) | Compound | Compound name | Compound InfoCard |
| Compound (KEGG ID only, e.g., C00020) | — | — | InfoCard created directly (not searchable) |
| Linked pathway (map entity) | — | — | New PathwayNode (direct navigation) |
| KO-only entity | Pathway | KO identifier | Pathway search results |
Cross-Pathway Navigation
When a pathway contains a link to another pathway (a "map" entity), double-clicking it loads that pathway directly as a new PathwayNode on the canvas. No command bar step is needed — it is direct navigation.
1. Load "mTOR signaling pathway"
2. See link to "PI3K-Akt signaling pathway" in the diagram
3. Double-click the linked pathway
4. PI3K-Akt pathway appears as a new node on the canvas
5. Both pathways are now visible and explorableThis lets you build out a multi-pathway canvas by following biological connections between signaling cascades.
Example Workflow: From Pathway to Protein Structure
This is the flagship use case — going from a systems-level pathway map to an atomic-level 3D structure in a few clicks.
-
Load a pathway — Search "mTOR signaling" in the command bar. A PathwayNode appears on the canvas.
-
Identify a gene of interest — Spot "PTEN" in the pathway diagram.
-
Double-click PTEN — The command bar opens with tag=Protein and query="PTEN".
-
Select the UniProt result — P60484 (PTEN, Homo sapiens) appears in the results.
-
View the 3D structure — A MolstarNode renders the PTEN crystal structure on the canvas.
-
Explore the structure — Rotate, zoom, measure distances, and change representation styles.
This pathway-to-structure pipeline lets you go from a systems-level view (pathway map) to atomic-level detail (3D protein structure) in seconds.
Special Cases
Bare KEGG Compound IDs
Some compounds in KEGG pathways only have identifiers like C00020 or C05981 with no human-readable name. These are not searchable in any database adapter. Instead, an InfoCard is created directly with the KEGG compound ID and a link to the KEGG compound page.
Undefined Entity Labels
Some KEGG pathway entries (typically group nodes) have no label in the KGML data. These appear without text in the pathway diagram. This is a cosmetic limitation of the KEGG data format and does not affect interaction with labeled entities.
Tips
Start broad, go deep. Load a disease pathway (e.g., "cancer") to get an overview, then double-click individual genes to explore their structures and functions.
Use pathway links. Many signaling pathways reference each other. Follow the links to build a multi-pathway canvas that maps out an entire biological system.
Combine with chat. After exploring a pathway, ask the AI: "Summarize the key genes in this pathway" or "Which of these genes are drug targets?"
Known Limitations
- KEGG compound IDs (e.g., C00020) are not searchable — they create InfoCards directly
- Tab completion selects only the first tag suggestion — use arrow keys to navigate alternatives
- Some KEGG group nodes have no label (data limitation from the KGML format)
- Pathway search works best with standard KEGG pathway names