> ## Documentation Index
> Fetch the complete documentation index at: https://docs.galagos.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Visualisations and reports

> Every Galagos project includes built-in viewers for sequences, structures, alignments, variants, trees, tables, and documents — plus auto-generated reports you can export to PDF, DOCX, or PPTX.

When you click a file in a project, Galagos picks the right viewer automatically. There is nothing to install, no separate window, and no download — the viewer opens inline next to the chat. When the agent produces a plot, the figure renders in the chat itself; ask for a different scale or a different split and the new figure replaces it within seconds.

## Built-in viewers

The viewer is determined by the file type. The same viewer drives both files you upload and files the agent produces.

> [card] 3D structure viewer
>
> PDB, mmCIF, SDF, MOL, MOL2, XYZ. Cartoon, stick, sphere, and surface representations. Rotate freely. Recolour by chain or secondary structure. Click two atoms to measure distance. Paste a PDB ID in the chat to fetch a structure directly from the database.

> [card] Sequence browser
>
> FASTA, FASTQ, GenBank. Scrollable sequence display with annotations. FASTQ files include quality strips. Long sequences load progressively.

> [card] Phylogenetic tree viewer
>
> Newick and NHX. Horizontal and radial layouts. Branch lengths preserved. Hover to see node and clade information.

> [card] Genome browser
>
> BAM, VCF, BED, GFF, GTF, BigWig. Standard track-based genome browser. Switch any track to a sortable table view with one click — useful for filtering variants or inspecting peak summits.

> [card] Variant and alignment tables
>
> VCF and BAM data as sortable, filterable tables. Pair with the genome browser to jump between the table and the locus.

> [card] Tabular viewer
>
> CSV and TSV files open as searchable, sortable, filterable tables. Click a column header to sort; click again to reverse.

> [card] Notebook and document viewer
>
> Jupyter notebooks render with code, output, and figures intact. Markdown, HTML, PDF, DOCX, and PPTX all open in place.

> [card] Image viewer
>
> PNG, JPG, SVG, GIF. Zoom and arrow-key navigation through multiple images. Useful for microscopy snapshots and figure drafts.

## Inline figures from the agent

Asking for "a volcano plot", "a heatmap", or "a UMAP" produces the figure rendered directly in the chat. The figure is interactive where appropriate — hover labels, axis scales, click-through to source rows — and downloadable as PNG or SVG from the icon in the figure's top-right corner.

Follow-up requests replace the existing figure rather than appending a new one:

* *"Now show that on a log scale."*
* *"Split the bars by condition."*
* *"Drop the unannotated genes."*
* *"Same plot but limit to the top 50 by adjusted p-value."*

The agent recomputes from the underlying data, not from the image, so the new figure is correct rather than re-rendered.

## Common figure types

| Figure | When the agent produces it |
| - | - |
| Volcano plot | Differential expression — fold change against significance, with thresholds annotated. |
| Heatmap | Expression clustering, correlation matrices, top-gene comparison. Hierarchical clustering by default. |
| UMAP, t-SNE | Single-cell embeddings. Colour by cluster, by condition, or by marker-gene expression on request. |
| Bar chart | Count comparisons, QC summaries, categorical breakdowns. |
| Boxplot, violin | Distribution comparisons across conditions or groups. |
| Manhattan plot | GWAS and genome-wide significance results. |
| ROC curve, PR curve | Classifier evaluation. |

Ask in plain English. *"A volcano plot, but colour upregulated genes in orange and downregulated in blue, with the top 20 labelled"* is a perfectly normal request.

> [!TIP]
>
> Vague prompts produce reasonable defaults. Specific prompts produce exactly what you ask for. *"Show the differential expression results"* gets you a volcano plot; *"a volcano plot of the DESeq2 results, log10 p-axis, labelled by gene symbol for any |log2FC| > 2"* gets you exactly that.

## Structured reports

When an analysis is non-trivial, the agent assembles a structured report alongside the figures. A report contains:

* **Analysis summary** — a plain-English description of the goal, the approach, and the tools used.
* **Parameters** — exact values passed to each tool (DESeq2 thresholds, GATK filters, Scanpy resolution).
* **Key findings** — statistical results, ranked gene lists, variant counts, cluster assignments.
* **Limitations paragraph** — caveats the agent flags about its own output (sample size, batch, normalisation, mapping coverage). See [Reproducibility and audit](https://docs.galagos.ai/analyses/reproducibility-and-audit.md).
* **Literature references** — citations from PubMed, UniProt, and NCBI relevant to the methods or findings.

Reports are designed to be readable top-to-bottom or skimmable directly to the parameters section.

## Exporting results

Every viewer and every report includes export controls. The right export format depends on what you are doing next:

> [card] PDF
>
> For reading and sharing. Markdown reports include a **Convert to PDF** button in the viewer. Useful for review packets and email.

> [card] DOCX
>
> For drafting manuscripts. Markdown reports include a **Convert to DOCX** button. Edit further in Word or Google Docs.

> [card] PPTX
>
> For talks. Ask the agent for a slide deck and you get a PPTX with figures, summaries, and conclusions already placed. Useful for lab meetings and conference talks.

> [card] PNG / SVG
>
> For individual figures. Download icon in the top-right corner of each plot.

You can also download an entire project as a ZIP — chat history, inputs, agent-written scripts, and outputs together. See [Reproducibility and audit](https://docs.galagos.ai/analyses/reproducibility-and-audit.md) for the full archive contents.

## A few tips

* **Click before you screenshot.** Most figures are interactive; you may not need the screenshot at all.
* **Ask, don't reformat.** If a plot is the right shape but the wrong colour or scale, say so. The agent re-renders from the data, not from the image.
* **Don't overshoot the format.** PDF for static sharing, DOCX for further writing, PPTX for talks. Pick the one closest to your next step.
* **Use the table-genome browser switch on VCFs.** Filtering hundreds of variants is easier in the table; spot-checking a single call is easier in the browser.

## Next steps

> [card] [Supported file formats](https://docs.galagos.ai/analyses/supported-file-formats.md)
>
> The full list of formats each viewer opens.

> [card] [Running an analysis](https://docs.galagos.ai/analyses/running-an-analysis.md)
>
> How to write prompts that produce the figures you need.

> [card] [Reproducibility and audit](https://docs.galagos.ai/analyses/reproducibility-and-audit.md)
>
> What gets recorded alongside the figures and the report.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.