MINERVA Documentation
A complete reference for the MINERVA platform — a knowledge-graph-based intelligence system for exploring disease–microbiome interactions.
Overview
MINERVA (Microbiome-disease INtEraction and Relationship Visualization & Analysis) is built on a corpus of 129,000+ peer-reviewed publications and provides four complementary analysis tracks for exploring how microbial communities relate to disease states.
All features are accessible from the top navigation bar and organised into three main modules:
Papers & Films
Explore the studies connected to MINERVA. Our research library brings together paper links, research summaries, and short films in one place.
MINERVA RESEARCHThe papers behind the platform.Browse papers, watch the films, and follow the original sources. Available without signing in.Browse papers & films →Quick Start
New to MINERVA? Follow these steps to get oriented in under 5 minutes.
Disease Analysis
Open →Search for any disease by name (UMLS CUI-mapped). Once selected, the panel renders three main sections:
| Graph-RAG Insight | LLM-generated headline and key findings grounded in the MINERVA corpus. Four preset queries: Summary, Key Microbes, Evidence Recency, Conflicts. A custom question input is also available. |
| Disease–Microbe Relations | Searchable, sortable table of all associated microbes ranked by association strength. Toggle All / Positive / Negative tabs. The first microbe is auto-selected on load; clicking any row opens a publication list filtered to that microbe. Advanced filter by JCR quartile (Q1–Q4) and impact factor. |
| Key Publications | Two sub-sections: a 10-year area chart (Recharts) showing annual publication volume, and a scrollable full publication list with PMID, journal, year, impact factor, and evidence type. |
How to use
- 1.Navigate to Entity Analysis → Disease Analysis.
- 2.Type a disease name in the search bar and select a result.
- 3.The panel auto-loads. Switch between Insight / Relations / Publications sections.
- 4.In the Insight panel, click a preset query or type a custom question.
- 5.In Relations, click any microbe row to see its specific publication evidence.
Microbe Analysis
Open →Identical layout to Disease Analysis, but the focal entity is a microbial taxon. Preset insight queries are: Summary, Key Diseases, Evidence Recency, and Conflicts.
How to use
- 1.Navigate to Entity Analysis → Microbe Analysis.
- 2.Search for a microbial taxon (e.g. Faecalibacterium prausnitzii, Lactobacillus).
- 3.Explore associated diseases, publications, trend chart, and LLM insight.
CrossLink Network Analysis
Open →CrossLink renders a bipartite force-directed graph. The anchor set is your chosen diseases or microbes; the opposite set is their linked counterparts. Designed for comparative questions like which microbes are dysregulated in both IBD and colorectal cancer? style comparative questions.
| Anchor Type | Choose whether anchors are diseases or microbes. |
| Anchor Selection | Search and add entities via CUI chips. Multiple anchors are supported. |
| Sign Filter | Show All / Positive (enriched) / Negative (depleted) associations only. |
| Strength Slider | Top-percentile cutoff — higher values show only the strongest edges. |
| Coverage Filter | Minimum number of anchors a node must connect to. Raise this to isolate shared (intersection) nodes. |
| Edge Detail Panel | Click any edge to open a right-side panel with its supporting publication list. |
How to use
- 1.Navigate to Graph Analysis.
- 2.Set Anchor Type (Disease or Microbe) and add 2–5 entities.
- 3.Adjust sign / strength / coverage filters to focus the graph.
- 4.Hover nodes for labels; click edges for supporting papers.
- 5.Highlighted intersection nodes appear in all selected anchors — prime candidates for shared dysbiotic taxa.
Introduction to Microbial Guilds
All four Guilds Analysis modules are built around a common framework of five mechanistic interaction axes. Each axis is derived from a distinct large-scale dataset:
| Axis | Name | Source Dataset |
|---|---|---|
| Axis 1 | Metabolic Exchange | Nature Communications 2023 — genome-scale metabolic models |
| Axis 2 | Spatial Co-occurrence | MicrobiomeNet 2025 — sample-level co-occurrence networks |
| Axis 3 | Antimicrobial Interaction | PairInteraX, Microbiome 2025 — pairwise inhibition/promotion scores |
| Axis 4 | Immune Modulation | Cell 2017 — Treg / Th17 / Th1 polarisation z-scores |
| Axis 5 | Taxonomic Relatedness | NCBI Taxonomy snapshot 2026-03-20 |
EIS Calculator
Open →The Ecological Interaction Score (EIS) quantifies the relationship between any two microbial species across all five axes into an interpretable score profile.
| Species Pair Input | Select Species A and B via autocomplete search. |
| 5-Axis Evidence Matrix | Per-axis signal badges — positive, negative, mixed, or unavailable — with a one-sentence interpretation headline for each axis. |
| Disease-Aware Layer | Optionally search for a disease to overlay its known dysbiosis context onto the pair's interaction profile. |
| Pair Sub-network Graph | Force graph showing the pair alongside their shared neighbors across active axes. |
| Edge Detail Panel | Click any edge for per-axis breakdown and underlying evidence. |
How to use
- 1.Navigate to Guilds Analysis → EIS Calculator.
- 2.Search and select Species A (e.g. Faecalibacterium prausnitzii).
- 3.Search and select Species B (e.g. Akkermansia muciniphila).
- 4.Review the 5-axis evidence matrix and signal badges.
- 5.Optionally search for a disease to activate the disease-aware interpretation layer.
Guild Graph
Open →Starting from one anchor species, the Guild Graph expands outward to show all species it interacts with across selected axes, forming a local guild network.
| Anchor Species | One species as the network focal point. |
| Species Recommendations | When selecting Species B, a dropdown shows recommended species that frequently co-appear with the anchor — sourced from the co-occurrence dataset. |
| Axis Selection | Toggle which of the 5 axes to include in the network expansion. |
| Neighbor Limit | Maximum number of neighbors pulled per axis. |
| Edge Colors & Signals | Each axis is color-coded. Signal values are positive, negative, mixed, or unavailable. |
How to use
- 1.Navigate to Guilds Analysis → Guild Graph.
- 2.Search and select one anchor species.
- 3.Toggle the axes you want to include.
- 4.Explore the force graph and click edges for axis-level breakdowns.
Disease-Guild Network
Open →The Disease-Guild Network integrates three layers — disease → microbes → guild interactions — into a single 3-layer force graph, revealing which microbial guilds are co-recruited by a disease's microbiome signature.
| Disease Search | Select a disease; MINERVA fetches its associated microbes from the knowledge graph. |
| Top Microbes Selector | Choose how many top-associated microbes to include (ranked by association strength). |
| Relation Type Filter | Restrict to positive, negative, or all disease–microbe associations. |
| Axis Filter | Include axes 1–4 (Metabolic, Spatial, Antimicrobial, Immune). Axis 5 (Taxonomic) is not used in this view. |
| Mapping Coverage | Shows what fraction of selected microbes were successfully mapped to guild interaction data. |
| Network Insights Panel | Structured panel showing hub microbes (high centrality), bridge species, detected communities, and indirect disease–disease overlap comparisons. |
| LLM Interpretation | Button-triggered feature that sends the network to AWS Bedrock and returns a free-form Markdown interpretation. |
How to use
- 1.Navigate to Guilds Analysis → Disease-Guild Network.
- 2.Search and select a disease.
- 3.Adjust Top Microbes, Relation Type, and Axis filters.
- 4.Click 'Apply Filters' to render the 3-layer graph.
- 5.Review the Network Insights panel for structured analysis.
- 6.Click the Interpret button to generate a free-form LLM summary.
Data Sources
MINERVA integrates several large-scale datasets. The table below maps each dataset to the features that depend on it.
| Dataset | Used in |
|---|---|
| MINERVA Knowledge Graph | Entity Analysis, CrossLink, Disease-Guild |
| Metabolic Exchange (Axis 1) & Spatial Co-occurrence (Axis 2) | EIS, Guild Graph, Disease-Guild |
| PairInteraX (Axis 3) | EIS, Guild Graph, Disease-Guild |
| Immune Modulation (Axis 4) | EIS, Guild Graph, Disease-Guild |
| NCBI Taxonomy (Axis 5) | EIS, Guild Graph |
Q&A / Contact
MINERVA includes a community Q&A system for researchers to ask questions about the platform, datasets, or findings.
| Browse Q&A | Visit the Q&A page to read all community questions and official team replies, ordered newest-first with a load-more option. |
| Submit a Question | Click 'Contact Us' in the home page footer to open the contact form modal. Provide your name, email, title, and message. |
| Reply Visibility | Official replies are marked with a badge. Each question carries a status: pending, answered, or closed. |