This document describes the output produced by the pipeline. The directories listed below will be created in the results directory after the pipeline has finished. All paths are relative to the top-level results directory.
The pipeline is built using Nextflow and produces the following output categories:
- Main outputs — Images and HTML files detailing major steps in the pipeline or analysis results
- Intermediate outputs — Files produced during workflow execution that are used further downstream
- Pipeline information — Report metrics generated during workflow execution
Tip
Example images of outputs are present in the Example outputs section.
Output files
<SAMPLENAME>/reports/- The HTML output from the R analysis step:
<SAMPLENAME>_report.html
- The HTML output from the R analysis step:
image_downscale/- If downscaling was performed, this directory will contain the downscaled TIFF:
<SAMPLENAME>.downscaled.ome.tif
- If downscaling was performed, this directory will contain the downscaled TIFF:
segmentation/- The segmentation mask output from Mesmer (default):
<SAMPLENAME>_mesmer.tif, or Cellpose:<SAMPLENAME>_cellpose.tif overviews/- Output segmentation image:
<SAMPLENAME>_bw_boundaries.tiff, which contains two channels — one with the processed DAPI, the other with the segmentation borders, both in black-and-white - Rendering of the segmentation borders (red) over the DAPI values (blue) in TIFF format:
<SAMPLENAME>_rgb_boundaries.tiff
- Output segmentation image:
<SEGMENTATION_METHOD>_mcquant/- The cell-by-feature matrix output from MCQuant:
<SAMPLENAME>.csv
- The cell-by-feature matrix output from MCQuant:
- The segmentation mask output from Mesmer (default):
seurat_objects/- The saved Seurat object, clustered, output from the R analysis step:
<SAMPLENAME>_seurat.rds
- The saved Seurat object, clustered, output from the R analysis step:
cluster_markers/- One CSV per clustering resolution (0.25, 0.5, 0.75, 1), containing mean intensity for each marker per cluster:
<SAMPLENAME>_<RESOLUTION>_cluster_mean_intensity.csv
- One CSV per clustering resolution (0.25, 0.5, 0.75, 1), containing mean intensity for each marker per cluster:
dapi_processed/- This directory will only be present if DAPI preprocessing was used in the pipeline.
- A TIFF image of the DAPI channel post background removal (if executed) and binarisation:
<SAMPLENAME>_dapi_processed.tif - A PNG image showing an overview of the DAPI channel pre- and post-processing, as well as a representative histogram with the Otsu threshold drawn:
<SAMPLENAME>_dapi_diagnostic.png
Output files
<SAMPLENAME>/extracted_channels/- The extracted DAPI channel used in segmentation:
<SAMPLENAME>_dapi.tif - If the AF subtraction method was used in DAPI preprocessing, the extracted channel will also be found here:
<SAMPLENAME>_AF.tif - Likewise, if a cell boundary stain was specified, this channel will also be present.
- The extracted DAPI channel used in segmentation:
metadata/- XML metadata extracted from the TIFF image:
<SAMPLENAME>.xml
- XML metadata extracted from the TIFF image:
image_hires/- This directory will contain the
<SAMPLENAME>.ome.tiffull-resolution image output from tile stitching or legacy HALO conversion — this is the input for the rest of the workflow.
- This directory will contain the
Output files
pipeline_info/- Reports generated by Nextflow:
execution_report.html,execution_timeline.html,execution_trace.txtandpipeline_dag.dot/pipeline_dag.svg. - Reports generated by the pipeline:
pipeline_report.html,pipeline_report.txtandsoftware_versions.yml. Thepipeline_report*files will only be present if the--email/--email_on_failparameters are used when running the pipeline. - Reformatted samplesheet files used as input to the pipeline:
samplesheet.valid.csv. - Parameters used by the pipeline run:
params.json.
- Reports generated by Nextflow:
Nextflow provides excellent functionality for generating various reports relevant to the running and execution of the pipeline. This will allow you to troubleshoot errors with the running of the pipeline, and also provides information such as launch commands, run times and resource usage.
Segmentation summary images
These outputs include the following two TIFF images (zoomed in here for clarity):
<SAMPLENAME>_bw_boundaries.tiff
| DAPI Channel | Segmentation boundaries |
|---|---|
 |
 |
<SAMPLENAME>_rgb_boundaries.tiff
DAPI preprocessing diagnostic
Apart from <SAMPLENAME>_dapi_processed.tif, which shows the output binarised/background-removed DAPI channel, the DAPI preprocessing step also outputs <SAMPLENAME>_dapi_diagnostic.png, an example of which is shown below.
This features, from left to right: the raw DAPI channel, the processed DAPI channel, and a histogram of DAPI intensities across the sample, with the Otsu threshold derived from the workflow drawn as a vertical line.
QC and clustering HTML output
In brief, the QC report outputs the following:
-
Tabular QC and statistical overviews for:
- Segmentation outputs (e.g. cell area, eccentricity, etc.)
- Inferred marker intensity from MCQuant (raw, log-transformed, and CLR-transformed)
Statistical summary — CLR transformed data
Marker Mean Median SD Min Max CRIP2 -0.3847963 -0.3721695 0.1047961 -0.9138822 0.6241850 DAPI 0.6234870 0.6084664 0.2371259 -0.2121824 1.6505124 HLA.ABC 0.2781894 0.2367254 0.3568910 -0.7763776 1.8521508 IRF9 -0.3403266 -0.3356826 0.0934302 -0.7458426 0.1718112 PANCK -0.2773508 -0.3112169 0.1768041 -0.8039722 0.9906032 PSTAT1 -0.3817753 -0.3797387 0.1076608 -0.8377921 0.5672849 YH2AX -0.1487883 -0.3007846 0.4199619 -0.8465718 2.3328215 -
Graphical overviews of marker intensity distribution, including boxplots and histograms
-
Further analysis on marker intensities, including marker correlation analysis and Quantile-Quantile plots
-
UMAP reductions of the CLR-transformed data, grouped by Louvain clustering output at various resolution levels (0.2, 0.3 ... 0.8)
-
Spatial representation of cells grouped by their Louvain clusters (at various resolutions)
-
Spatial representation of cells coloured by their marker intensity levels
