> For the complete documentation index, see [llms.txt](https://docs.mithrl.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.mithrl.com/docs/mithrl-knowledge-graph.md).

# Mithrl Knowledge Graph

The Mithrl-1 Knowledge Graph (MKG) is not a generic graph store; it is a causally structured map of biology. Every relationship in the graph carries directionality (upstream cause to downstream effect), a defined interaction type (activating or inhibiting, direct or inferred), and traceable provenance back to its source literature, rather than an unlabeled edge saying two entities are merely connected. That typed, directional structure is what supports Mithrl-1's causal analytics, including upstream regulator network prediction, causal and mechanistic network construction, and toxicity function scoring, because each of those depends on knowing which entity drives which, how confidently, and whether the underlying evidence agrees or conflicts. The platform is built on a scalable, stable, and enforceable ontology that makes governance easy with every data point traced to verified sources like Cell, Reactome, PubChem and more. Most graph databases are infrastructure agnostic to the content sitting on top of them; the Mithrl-1 Knowledge Graph's differentiator is that the biological meaning, direction, and evidence weight are baked into the edges themselves before a query ever runs.

<figure><img src="https://626853399-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWzmerQ3GQmDagWwm1Oeg%2Fuploads%2FmWnNaQwpZeTkmocttEEf%2FUntitled%20(2).png?alt=media&amp;token=5e834896-31d8-4451-ba76-1acd17db4af7" alt=""><figcaption></figcaption></figure>

The MKG is built from aggregating \~[37 public and private data sources](/docs/data-sources.md) into a single graph store. The resultant Knowledge Graph contains:

* \~7 million nodes
* \~39 million edges

To traverse the MKG, there are several inference algorithms available. Documentation can be found in the Mithrl[-1 Inference Engines](/docs/mithrl-1-inference-engines.md).