Mithrl-1 Inference Engines

The following are built-in inference algorithms used to navigate the Mithrl Knowledge Graph.

Name	Short Description	Scientific Question
Pathway Participants	For one named pathway, list its participants in a chosen ontological category with the provenance of every membership assertion, plus a diagram reference derived from the pathway's CURIE.	What is actually in the pathway I named — which entities does this build assert participate in it, on whose authority, and where can I see its diagram?
Compound Tox Profile Card	For one compound CURIE, list the diseases CTD curators directly associated with it, labelled marker/mechanism or therapeutic, with the citations behind each association.	Which diseases has CTD directly curated for the compound I named, on which DirectEvidence, and on what published evidence — separately from anything merely inferred through its affected genes?
Upstream Regulators	Find the upstream regulators for a set of genes/features.	Which regulators are most likely driving the supplied expression signature?
Upstream Regulator Analysis	Test which upstream regulators are over-represented among the features you measured, against an explicit assayed background, and return each hit with its overlapping targets and citations.	Given a feature table, which upstream regulators have significantly more of their known targets among my features than the measured background would predict?
Toxicity Endpoint Analysis	Test which of the 26 curated toxicity endpoints are over-represented among the features you measured, against an explicit assayed background, with citations on every published endpoint.	Given a feature table, which curated toxicity endpoints are enriched for my features relative to the measured background — i.e. what toxicity is this signature consistent with?
Regulator Breakdown	For the regulators you name, return every asserted regulator-target edge with its predicate, qualifiers, sign and citations — uncapped and unaggregated.	For the regulators I named, which targets does each act on, by what predicate, with what polarity and on which publications — the per-edge evidence an enrichment result summarises away?
Pathway Lookup	List every annotated Reactome pathway the supplied genes participate in, and which genes fall in each — membership only, with no statistical test.	Which Reactome pathways contain the genes I supplied, and which of my genes are in each, including the pathways an enrichment test would drop for being too small or unremarkable?
Over-representation	Test a feature list for over-representation in a supplied term family.	Which terms are over-represented in this feature list, against an explicit background?
Compound Tox Profile	Needs one upload — omics_csv, a feature table whose entity column holds your compounds (rename the column with --param entity_column=<name>). Lists the genes each compound is asserted to affect with the sign and citations of every edge, alongside the member genes of the 26 curated toxicity endpoints.	For the compounds I named, which genes does each one affect and on what evidence, and how do those targets sit against the curated toxicity endpoint gene sets?
Canonical Pathway Analysis	Needs TWO uploads — omics_csv (your feature table) and background_csv (the assayed background). Tests which annotated pathways are over-represented among the features you measured against that background, returning each hit with its overlapping members and citations. If you have no background to declare, `pathway-lookup` lists the pathways containing your genes from omics_csv alone, without a statistical test.	Given a feature table, which canonical pathways contain significantly more of my features than the measured background would predict?