Visium HD integration

[ch99l]

No description provided.

[Copilot]

Pull request overview

This PR adds a dedicated Visium HD execution path to the Nextflow pipeline, alongside a refactor of standard clustering/quantification modules to support both automated and “manual clustering” restarts. It also introduces separate bambu and seurat containers and expands smoke tests to cover the new workflows.

Changes:

• Added a Visium HD workflow (--visium_hd) including bin aggregation, spot/bin mapping, optional Banksy clustering, and per-resolution quantification.
• Refactored standard clustering + quantification into shared/standard/visium_hd module structure; renamed --resolution → --seurat_resolution and EM_clusters → clusteredEM.
• Added --manual_clustering restart mode, updated docs/examples, and extended CI smoke tests + container build workflow.

Reviewed changes

Copilot reviewed 38 out of 60 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
subworkflows/prepare_input_visium_hd.nf	New Visium HD input preparation subworkflow.
subworkflows/prepare_input_standard.nf	Updated module include paths for standard input prep.
subworkflows/clustering.nf	Removed legacy clustering subworkflow.
subworkflows/clustering_standard.nf	New standard clustering subworkflow with updated inputs.
README.md	Documented Visium HD + manual clustering + parameter renames.
nextflow.config	Added Visium HD/manual params; renamed clustering params.
modules/seurat/visium_hd/clustering.nf	New Seurat/Banksy clustering process for Visium HD.
modules/seurat/standard/single_sample_clustering.nf	Refactored single-sample clustering to use count dirs + colData.
modules/seurat/standard/multi_sample_clustering.nf	Refactored multi-sample clustering (Harmony integration) inputs.
modules/prepare_input/visium_hd/spot_bin_mappings.nf	New process to produce spot→bin mapping CSVs.
modules/prepare_input/visium_hd/filter_barcoded_bam.nf	New BAM filter process using in-tissue barcodes.
modules/prepare_input/visium_hd/convert_tissue_positions.nf	New parquet→CSV conversion for tissue positions (+ barcodes).
modules/prepare_input/visium_hd/convert_barcode_mappings.nf	New parquet→CSV conversion for barcode mappings per resolution.
modules/prepare_input/standard/extract_spatial_coordinates.nf	New process extracting spatial coordinates for Visium chemistries.
modules/prepare_input/standard/extract_barcodes.nf	New process extracting 10x barcode whitelist per chemistry.
modules/bambu/visium_hd/transcript_discovery.nf	New transcript discovery for Visium HD at 2µm.
modules/bambu/visium_hd/spot_level_quantification.nf	New per-resolution spot/bin-level EM quantification for Visium HD.
modules/bambu/visium_hd/map_clusters_to_2um_spots.nf	New mapping of bin clusters down to 2µm spots.
modules/bambu/visium_hd/extract_spot_barcodes.nf	New helper to extract 2µm barcodes + sample name from quantData.
modules/bambu/visium_hd/aggregate_bins.nf	New aggregation of 2µm SE into coarser bin resolutions.
modules/bambu/standard/transcript_discovery.nf	Refactored standard discovery; now always publishes restart artifacts.
modules/bambu/standard/single_cell_quantification.nf	New single-cell EM quantification module (standard workflow).
modules/bambu/shared/prepare_annotation.nf	Retagged process label to bambu container.
modules/bambu/shared/construct_read_class.nf	Retagged process label to bambu container.
modules/bambu/shared/cluster_level_quantification.nf	Refactored cluster-level quantification into shared module.
main.nf	Split into STANDARD and VISIUM_HD workflows; rewired modules.
lib/Validation.groovy	Added Visium HD + new param validations; renamed seurat param.
examples/samplesheet_visium_hd_synthetic.csv	Added Visium HD synthetic example samplesheet.
examples/samplesheet_visium_hd_quant_data.csv	Added Visium HD manual-clustering restart samplesheet.
examples/samplesheet_test_sc_quant_data.csv	Added standard manual-clustering restart samplesheet.
examples/bins_visium_hd_synthetic.csv	Added synthetic Visium HD bins CSV example.
containers/seurat/Dockerfile	New Seurat container (Seurat, Harmony, Banksy, wrappers).
containers/bambu/Dockerfile	Updated bambu container dependencies and removed Seurat/Harmony.
conf/smoke_test.config	Added smoke-test profiles for Visium HD + manual clustering.
conf/dev.config	Updated valid quantification mode list (clusteredEM).
conf/containers.config	Split r into bambu + seurat; add pyarrow/pandas image.
conf/base.config	Updated retry strategy labels to `bambu
CHANGELOG.md	Added v0.10.0 notes for Visium HD + refactor.
bin/visium_hd_spot_bin_mappings.py	New script to subset barcode mappings to quantified spots.
bin/visium_hd_convert_tissue_positions.py	New script to convert tissue_positions parquet→CSV (+ barcodes).
bin/visium_hd_convert_barcode_mappings.py	New script to subset barcode_mappings parquet per resolution.
bin/visium_hd_cluster.R	New R helpers for Banksy/expression-only Visium HD clustering.
bin/visium_hd_aggregate_resolution.R	New R helper to aggregate 2µm counts to bins.
bin/save_counts.R	Renamed helper to saveCounts() and refactored variable names.
bin/map_spots_to_clusters.R	New R helpers to read clusters and expand bin clusters to spots.
bin/create_seurat_object.R	New R helper to build Seurat objects from count directories.
bin/bambu_discovery.R	New shared R helper encapsulating discovery + quantData + uniqueCounts.
.github/workflows/smoke_test.yml	Expanded smoke-test matrix with Visium HD/manual variants.
.github/workflows/build_container.yml	Generalized container build to multiple labeled images.

Suppressed comments (1)

modules/bambu/shared/cluster_level_quantification.nf:31

• Manual clustering docs indicate the clusters input can be a CSV/TSV with id,cluster, but this code passes non-.rds inputs as a string path into bambu.singlecell(). That’s unlikely to match the expected named-vector type and will break manual clustering with CSV input. Parse the file into an id -> cluster named vector (the repo already includes a readClusters() helper).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.