# Analysis Modules

This section is the operational reference for each EVd3x analysis lens.

## Pathway Analysis

### What it does
- Runs enrichment-style interpretation for the current node or collection context.
- Organizes results by source and category so you can compare KEGG, Reactome, GO, and WikiPathways consistently.

### Inputs
- Active genes from the current graph/session context.
- Pathway memberships (`gene_pathways.parquet`).
- Active filters (p-value threshold, keyword, source toggles).

### Outputs
- Filtered pathway list for fast UI interpretation.
- Full pathway export tables for downstream analysis.

### Interpretation notes
- Compare pathways inside the same filter state.
- Use q-value/p-value in context; do not compare directly to disease or communication scores.

## Disease Analysis

### What it does
- Aggregates disease evidence by canonical disease identity and display-normalized labels.
- Preserves evidence depth signals and supporting entity counts.

### Inputs
- Disease-linked evidence rows from gene/miRNA context.
- Canonical grouping keys (`Disease_ID`).

### Outputs
- Grouped disease summary cards and filtered views.
- Full disease association exports with source/publication fields.

### Interpretation notes
- Treat as association context, not causality.
- Check supporting entity diversity and source depth before prioritization.

## Cell Types (Cell Specificity)

### What it does
- Scores likely cell-context relevance for current molecular context.
- Combines marker/context evidence for ranked cell outputs.

### Inputs
- Cell specificity tables and expression-linked context.
- Active query context and top-N scope.

### Outputs
- Ranked cell list and system-level context views.
- Export-ready ranking tables.

### Interpretation notes
- Cell ranking is context-dependent and query-sensitive.
- Use with EV Evidence and Pathway to avoid over-interpretation.

## Cell Communication / L-R Pairs

### What it does
- Uses ligand-receptor mapping and bridge context to expose communication hypotheses.
- Supports pair-focused drilldown in active selected cell context.

### Inputs
- Ligand/receptor resources (`ligand_receptor_pairs_full.parquet`).
- Selected cells and active context.

### Outputs
- Pair rows, summary matrices/heatmaps, and selected-pair detail views.
- Full interaction exports.

### Interpretation notes
- Communication views may be capped in UI for speed; use full exports when needed.
- Validate with known biology and source provenance.

## EV Evidence

### What it does
- Shows EV database support and publication/provenance context for current entities.
- Surfaces EV-TRACK fields where present in upstream evidence rows.

### Inputs
- EV evidence tables (`ev_evidence.parquet`) and linked publication context.

### Outputs
- EV evidence row view in UI.
- Full source-aware EV evidence exports.

### Interpretation notes
- Missing EV-TRACK ID can reflect source availability, not necessarily absence of EV relevance.
- Prioritize rows with stronger provenance fields for reporting.

## PPI

### What it does
- Builds protein interaction neighborhoods for selected proteins.
- Supports thresholded one-hop exploration in the right panel and graph overlays.

### Inputs
- STRING-derived interactions and active threshold settings.

### Outputs
- PPI tree panel view and export rows.
- Optional graph overlays for partner-focused exploration.

### Interpretation notes
- PPI is context support, not direct pathway causation.
- Report active threshold and view cap when sharing results.

## Recommended reading order by use case

- Fast interpretation: Summary -> EV Evidence -> Pathway -> Disease.
- Mechanistic follow-up: Cell Types -> L-R -> PPI.
- Reporting/export: module tab view -> full export tables.