Links each curated study record back to its DOI, letting the research assistant return persistent, citable identifiers for the primary sources behind its answers.
Draws its 400+ hand-curated mTOR studies from PubMed records, so answers from the citation-grounded research assistant can be traced back to the underlying PubMed entries.
Uses the Zenodo record (DOI 10.5281/zenodo.22059963) as the archival, citable home of the curated dataset, providing identifier and citation information for the corpus.
This server has a remotely hosted endpoint. Open its connector listing for connection details, authentication requirements, and setup instructions:
A curated, evidence-graded database of mTOR pathway research in which every claim carries its source, the conditions it was measured under, and the point where it stops holding. Studies are labelled by the kind of study behind them - from synthesis of human data down to mechanistic and in-vitro work - and traced back to their primary source, alongside a knowledge-graph view of genes, diseases, and interventions and a layer of open questions naming what the evidence does not yet resolve.
Live site: https://mtor-atlas.org
400+ hand-curated primary studies on the mTOR signaling pathway (mTORC1/mTORC2, autophagy, rapamycin and related interventions), each labelled by the kind of study behind it and linked back to its DOI/PubMed record.
A knowledge-graph view connecting genes, diseases, and interventions.
An "open questions" layer - evidence gaps identified across the corpus, each paired with a proposed testable experiment.
A citation-grounded research assistant that answers pathway questions using only the indexed corpus, with links back to source studies.
Related MCP server: utopia-mcp
Studies are hand-selected from PubMed / Europe PMC and labelled by study design, not by quality, importance, or citation count:
S - synthesis of human data (systematic review / meta-analysis)
H - human study (clinical trial or observational)
A - animal model
M - molecular / in-vitro (mechanistic)
R - review
These codes ran A-D until September 2026. They were renamed because a lettered ladder reads as a quality grade, which it never was, and because the old bottom tier merged primary mechanistic work with narrative reviews - two different kinds of claim. The change was prompted by an external critique from a researcher in the field; the underlying data was not re-graded, only the labels shown to readers.
A mechanistic paper is not "worse" than a trial. The code says what kind of claim a study can support, not how good it is.
Built and maintained independently by Oliver, a high-school student, together with his father Petr. Not affiliated with any lab, company, or institution. Feedback on the evidence grading, missing studies, or anything that looks wrong is very welcome - please open an issue. See CONTRIBUTING.md.
JSON API (read-only, no key): https://mtor-atlas.org/api/ - studies with evidence codes, entities, signed pathway relations with supporting and conflicting studies, open questions. OpenAPI 3.1: https://mtor-atlas.org/api/openapi.json
MCP server for AI assistants: source and install instructions in mcp/.
If you use this dataset, please cite it via its Zenodo record: https://doi.org/10.5281/zenodo.22059963
A single page with all identifiers, registrations (bio.tools, FAIRsharing, GitHub, ORCID) and a ready-to-use citation is at https://mtor-atlas.org/data/.
This repository is dual-licensed, because it contains two different kinds of thing:
Curated content and data - the study records, evidence grades, curated prose, gap hypotheses, and everything under atlas_data/ and the generated pages - are licensed under CC BY 4.0 (see LICENSE): https://creativecommons.org/licenses/by/4.0/
Source code - the Python generators, validation and verification scripts, and site JavaScript - is licensed under the MIT License (see LICENSE-CODE).
If you reuse the data, attribute it. If you reuse the code, MIT terms apply.
Dataset version, corpus snapshot date, counts, evidence-code legend and how to cite the dataset.
Inputs: no named parameters.
Direct curated relations between two entities (either direction) with full cards for the supporting and conflicting studies. Answers questions such as "what is the evidence that A acts on B?".
Inputs: a (required), b (required).
Relations the Atlas marks as contested or that carry conflicting studies, with both sides of the evidence. Optionally limited to one entity.
Inputs: limit, entity.
Signed, evidence-linked pathway relations (claims such as "Rheb activates mTORC1"). Filter by an entity on either end, by source/target, effect, contested status or minimum strength of the best supporting evidence.
Inputs: limit, effect, entity, offset, +4 more.
One entity (gene/protein, complex, drug, disease, process...) with its linked studies as short cards (first studies_limit) and every pathway relation it takes part in as a one-line claim. Accepts an id, a name or a synonym.
Inputs: entity (required), studies_limit.
One open or frontier question by ID (e.g. "H1", "F2"): the gap, what changed, what is still open, how it could be tested, linked studies.
Inputs: id (required).
One pathway relation by ID (e.g. "RHEB-MTORC1"): mechanism, boundary conditions, confidence, supporting and conflicting studies.
Inputs: id (required).
Full record for one study by Atlas ID (e.g. "SAB1994"): abstract excerpt, extracted findings, linked entities, the relations it supports or contradicts, related open questions.
Inputs: sid (required).
The Atlas's open questions (evidence gaps with testable hypotheses) and frontier questions.
Inputs: kind.
Find genes/proteins, complexes, drugs, diseases, processes, nutrients and outcomes by name or synonym.
Inputs: query (required), type, limit.
Search the curated studies by keywords (title, finding, authors, journal, model system), optionally filtered by evidence code, a linked entity (gene, drug, disease...) and year range. Returns summary records with evidence codes and URLs.
Inputs: limit, query, entity, offset, +3 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 11 tools
Most tools target distinct resources (studies, entities, relations, questions) with clear get_/search_ pairs. However, evidence_between overlaps with find_relations (both return curated relations, the former just a special case for a two-entity pair), and find_contradictions overlaps with find_relations' contested-status filter. Descriptions help but an agent could still misselect among the relation-oriented tools.
The set is predominantly snake_case verb_noun (search_studies, get_study, find_relations, list_questions), which is readable and predictable. The main deviation is the inconsistent use of search_ vs find_ for essentially the same lookup semantics, plus the noun-only atlas_about.
11 tools is well-scoped for a read-only curated atlas, with each tool mapping to a distinct resource or query pattern. No tool feels redundant or trivially thin.
The surface covers the full read lifecycle for the domain: metadata (atlas_about), search+get for studies, entities, relations and questions, plus specialized views for evidence-between and contradictions. As a curated read-only dataset, no create/update/delete is needed, so coverage is complete.
Read-only Kyntrax definitions, claims, evidence status and public provenance.
Public read-only MCP access to digital learning concepts, semantic relationships, and citations.
Read-only cannabis compound, receptor, anatomy and evidence data, every value sourced.
Read-only access to approved maxand98 publications, citations, claims and evidence records.
