nf-core pipeline · fish, image-processing, imaging

nf-core/molkart

An automated pipeline for processing Resolve Biosciences Molecular Cartography combinatorial FISH data.

Run it with the research agent Documentation Source code

What nf-core/molkart is

nf-core/molkart is a bioinformatics pipeline designed for end-to-end processing of Molecular Cartography spatial transcriptomics data generated on the Resolve Biosciences platform. Originally written by Kresimir Bestak and Florian Wuennemann, the workflow handles data generated by combinatorial fluorescence in situ hybridisation (FISH).

Spatial transcriptomics experiments yield microscopic image tiles alongside three-dimensional coordinates of detected fluorescent transcripts. To extract biological insights, researchers must clean imaging artefacts such as grid seams, segment individual cells, and assign transcript coordinates to their corresponding cellular boundaries. nf-core/molkart automates these computational steps into a reproducible workflow.

The pipeline produces a cell-by-transcript expression matrix from deduplicated spatial coordinates and segmented boundaries. It also generates diagnostic metrics, including transcript assignment efficiency and cell size distributions, integrated into a summary quality-control report.

What the pipeline does

The workflow begins by accepting a field-of-view samplesheet containing a tabular list of FISH spot coordinates (x, y, z, and gene identity), a nuclear DAPI image in TIFF format, and an optional membrane or secondary staining TIFF image. Image preprocessing employs MindaGap to fill grid patterns arising from imaging tiles. Users can optionally apply contrast-limited adaptive histogram equalisation, and merge multi-stain inputs into multichannel image stacks.

For cellular boundary detection, the workflow supports multiple segmentation algorithms: Cellpose, Mesmer, ilastik, or StarDist. Segmented objects are filtered by size thresholds to eliminate non-cellular artefacts. Concurrently, MindaGap detects and removes duplicated spot coordinates situated near image grid seams. Finally, deduplicated transcripts are mapped into cellular masks to generate a cell-by-gene count matrix, and quality metrics are compiled into an interactive MultiQC report.

Key facts

Data type: Molecular Cartography combinatorial FISH
Original developers: Kresimir Bestak and Florian Wuennemann
Primary inputs: TIFF images (DAPI and optional membrane stain) and spot coordinate tables (x, y, z, gene)
Primary outputs: Cell-by-transcript expression matrix, segmentation masks, and MultiQC report
Supported segmenters: Cellpose, Mesmer, ilastik, and StarDist
Image artefact correction: MindaGap
Workflow framework: Nextflow (nf-core)
Release version: 1.2.0

What you can use it for

Molecular Cartography data processing: Execute end-to-end processing of Resolve Biosciences combinatorial FISH datasets from raw DAPI images and spot tables to single-cell count matrices.
Multi-algorithm cell segmentation: Segment cell boundaries across fields of view using Cellpose, Mesmer, ilastik, or StarDist with single-channel or multi-channel imaging data.
Image grid artefact removal: Eliminate image seam lines and deduplicate artificial transcript spots appearing along grid boundaries using integrated MindaGap functionality.
Transcriptomic quality assessment: Evaluate spatial assay performance by inspecting spot assignment rates, average transcript count per cell, and cell size distributions within a MultiQC report.

Run it here

The research agent runs nf-core/molkart with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.

Ask the research agent

Strengths and limits

Good at

Supports four different cell segmentation engines: Cellpose, Mesmer, ilastik, and StarDist.
Directly resolves tiling artefacts by filling grid lines and deduplicating nearby spots with MindaGap.
Produces standardised cell-by-transcript matrices alongside intermediary segmentation masks for downstream spatial analysis.
Compiles pipeline-specific quality metrics into MultiQC reports to evaluate cell-assignment rates and mask sizes.

Watch out for

Specifically tailored to Resolve Biosciences Molecular Cartography data and expects predetermined spot table columns (x, y, z, gene).
Requires input images to be supplied in TIFF format.
Depends on predefined nuclear staining (DAPI) and supports only one additional optional stain (such as a membrane marker) for segmentation.

Questions

Can I run nf-core/molkart without my own cluster?

What licence is nf-core/molkart under?

MIT, like every nf-core pipeline.

What input file formats are required by the pipeline?

The pipeline requires a CSV samplesheet referencing a DAPI nuclear image in TIFF format, a tabular text file containing transcript spot coordinates with columns for x, y, z, and gene, and optionally a secondary TIFF image such as a membrane stain.

Which segmentation tools are available in nf-core/molkart?

Users can select between Cellpose, Mesmer, ilastik, and StarDist for cell segmentation.

How does the pipeline handle image tiling artefacts?

It applies MindaGap to fill tile grid patterns in the images and identify duplicate transcript spots located near the grid seams.

What quality metrics are generated during the analysis?

The pipeline calculates metrics including the spot assignment rate, average spots per segmented cell, and the distribution of segmentation mask sizes, which are summarised using MultiQC.

How to cite it

doi: 10.5281/zenodo.10650749

Written from nf-core/molkart's own README. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.