---
title: Scanpy outputs
description: 'Read the processed .h5ad, UMAP coordinates CSV, cluster assignments and per-cluster marker genes ranked by log fold change and adjusted p-value.'
---

You receive a processed AnnData file with UMAP embeddings, cluster assignments, and per-cluster marker genes, plus publication-ready figures. These let you annotate cell types from the top markers, feed the .h5ad into scvi-tools for probabilistic modeling, and identify which cell types express a target of interest.

## Output files

## What you receive

- **Processed AnnData**: .h5ad file with all computed embeddings and clusters
- **UMAP coordinates**: CSV with cell embeddings
- **Cluster assignments**: Cell-to-cluster mapping table
- **Marker genes**: Top differentially expressed genes per cluster
- **Figures**: Publication-ready UMAP and violin plots

## Interpreting results

### UMAP visualization

- Each point represents a cell
- Colors indicate cluster assignment
- Proximity suggests transcriptional similarity

### Marker genes

- Ranked by log fold change and adjusted p-value
- Use top markers to annotate cell types
- Cross-reference with known cell type markers

## Export formats

- **CSV**: Tabular data for downstream analysis
- **H5AD**: Full AnnData for scvi-tools or other Python tools
- **PNG/SVG**: Figures for publications

## Canvas Visualization

When a Scanpy job completes, a result card appears on the canvas.

- **Card size**: 450x350px
- **Border color**: Brown (#795548)
- **Visualization**: UMAP/t-SNE dimensionality reduction plots with cluster coloring

### Tier 2 Metrics

- Cell count
- Gene count
- Cluster count
- Median gene count per cell

### QC Thresholds

| Parameter | Default | Purpose |
|-----------|---------|---------|
| Min genes/cell | 200 | Filters out empty droplets |
| Min cells/gene | 3 | Removes very rare genes |
| Max mito % | 5 | Filters dying cells |
| Resolution | 0.5 | Controls cluster granularity |

## Read next