scpca-nf v0.8.5

What's Changed

• Use scpcaTools 0.4.1 images, which gives two major updates:
  • raw single-cell counts are rounded (as they were meant to be)
  • security updates with updated lightning in python packages that depend on it
• converted use of deprecated addParams() to take for future Nextflow compatibility
• correctly handle 2 cell case which does allow PCA creation

Full Changelog: v0.8.4...v0.8.5

scpca-nf v0.8.4

What's Changed

Bug fix release:

• Only count single cell libraries for max merged
• Account for different HVG name in merged object

Full Changelog: v0.8.3...v0.8.4

scpca-nf v0.8.3

What's Changed

• Updates to AnnData objects:

  • PCA and UMAP embeddings are now stored as X_pca and X_umap (previously X_PCA and X_UMAP)
  • .var contains a highly_variable column indicating which genes are part of the list of highly variable genes
  • .uns contains a pca object with parameters and variance weights associated with the PCA embeddings

• Addition of a max_merge_libraries parameter for merge.nf to limit the number of libraries that can be merged into a single object.

Full Changelog: v0.8.2...v0.8.3

scpca-nf v0.8.2

What's Changed

• Use new, slimmer scpcaTools docker containers (see https://github.com/AlexsLemonade/scpcaTools/releases/tag/v0.4.0)
• Bulk files, bulk_quant.tsv and bulk_metadata.tsv, are now saved to a bulk directory
• Main QC reports and cell type reports now include a table summarizing sample metadata (age, sex, diagnosis, etc.)

Full Changelog: v0.8.1...v0.8.2

scpca-nf v0.8.1

What's Changed

• Output HDF5 files now have the .h5ad extension instead of .hdf5
• Bug fixes:
  • Skip running CellAssign if a processed object only has 1 cell
  • No cell type reports are created if processed objects only have 1 cell
  • Make sure to account for no genetic demultiplexing when estimating sample cell counts for metadata.json
  • Skip any downstream processing if no cells are left after filtering out empty droplets

Full Changelog: v0.8.0...v0.8.1

scpca-nf v0.8.0

What's Changed

• Addition of merge.nf used to create merged SingleCellExperiment and AnnData objects
  • The merged workflow takes as input a list of projects and creates a single merged object for each project
  • Also includes a report, summarizing the contents of the merged object
• Some minor aesthetic fixes to cell type report
• Use bz2 compression to reduce file sizes for SingleCellExperiment objects
• Add nextflow_schema.json to repo

Full Changelog: v0.7.3...v0.8.0

scpca-nf v0.7.3

What's Changed

• Skip any additional processing (clustering, cell type assignment) for processed objects with 0 cells
• Use new max_mem label for CellAssign classification, which uses params.max_memory after the first failure.

Full Changelog: v0.7.2...v0.7.3

scpca-nf v0.7.2

What's Changed

• Skip CellAssign if too few cells
• Small bug fixes for running multiplexed samples through cell typing
• Skip cell typing for cell line samples
• Small adjustments to UMAPs in the cell typing report
• Account for missing input files

Full Changelog: v0.7.1...v0.7.2

scpca-nf v0.7.1

What's Changed

• Update memory and CPU requests for running CellAssign
• Set max retries after AWS instance failure to 2
• Use On Demand instances on 3rd retry
• Create faceted UMAPs in cell type report

Full Changelog: v0.7.0...v0.7.1

scpca-nf v0.7.0

What's Changed

• Addition of cell type annotation to the main workflow, including:
  • Annotation with either SingleR or CellAssign
  • New output that includes a processed SingleCellExperiment and AnnData objects with cell type annotations and a cell type summary report
  • Documentation for adding cell type annotations
• By default, cell type annotation is turned off. To run the workflow with cell type annotation, use the --perform_celltyping flag.
• By default, if cell typing results already exist, previous results are used. The --repeat_celltyping flag can be used to repeat previous cell type annotation.
• Addition of pre-commit to the repo

Full Changelog: v0.6.3...v0.7.0