Skip to content

Files

Latest commit

 

History

History
209 lines (92 loc) · 10.1 KB

README.md

File metadata and controls

209 lines (92 loc) · 10.1 KB

Description of the pipeline parameters

To find out the default value for each parameter, see conf/analysis.config.

Sequencing analysis parameters

  • do_merge_mtx Merge graft and host MTX (gene by spot) matrices into one MTX matrix

  • do_splicing_quantification Run splicing quantification with velocyto. The pipeline also sorts by cell barcodes the BAM file produced by Space Ranger.

  • do_snv_extract Run the BAF extraction sub-workflow to get bulk-level SNV.

  • reference_genome Path to the reference genome to use for Space Ranger reads alignment in one-reference analysis route. See https://support.10xgenomics.com/single-cell-gene-expression/software/release-notes/build for Space Ranger requirements of the reference genomes.

  • mouse_reference_genome Path to the mouse reference genome for Space Ranger reads alignment in two-reference analysis route.

  • human_reference_genome Path to the human reference genome for Space Ranger reads alignment in two-reference analysis route.

  • deconvolution_reference_graft Path to a graft (e.g., human) reference genome (e.g., *.fa, *.fna, *.fa.gz, *.fna.gz) to build xenome or xengsort indices. If the indices supplied in nextflow.config already exits, then this parameter is ignored.

  • deconvolution_reference_host Path to a host (e.g., mouse) reference genome (e.g., *.fa, *.fna, *.fa.gz, *.fna.gz) to build xenome or xengsort indices. If the indices supplied in nextflow.config already exits, then this parameter is ignored.

  • deconvolution_kmer_size K-mer size for building xenome or xengsort indices. See https://github.com/data61/gossamer/blob/master/docs/xenome.md for a detailed description.

  • deconvolution_indices_path Path to save deconvolution indices.

  • deconvolution_indices_name Name of the indices.

  • xengsort_n Xengsort-specific parameter. See https://gitlab.com/genomeinformatics/xengsort for details.

See https://github.com/akdess/BAFExtract for the description of the following filtering parameters:

  • bafextract_minimum_mapping_quality

  • bafextract_minimum_base_quality

  • bafextract_min_coverage_per_SNV

  • bafextract_min_MAF_covg_per_SNV

  • bafextract_min_MAF

Imaging analysis parameters

  • do_img_subworkflow Run the imaging sub-workflow to generate imaging and nuclear morphometric features for each spot on the grid.

  • short_workflow Run short imaging workflow instead of the full imaging workflow. See config for details.

  • do_imaging_anndata Create an AnnData object (e.g., for use with Scanpy) from the *.csv.gz data file with imaging and nuclear morphometric features

  • do_nuclear_sementation Perform nuclear segmentation (use either HoVer-Net or StarDist to segment nuclei) of the entire WSI.

  • target_mpp desired image resolution for scaling the images. Note that specific DL and ML models require full-resolution images, and the supplied pre-trained models are designed for images with a resolution of around 0.25 (mpp). In case a low-magnification image is supplied (e.g., mpp is 0.5) while target_mpp is 0.25, the image is upsampled and will have doubled dimensions.

  • tiled_tiff_tile_size The TIFF WSI is internally stored in blocks (for memory management). The tile size determines the block size. This parameter is not the size of tiles used for feature extraction or segmentation aggregation. The grid parameter grid_spot_diamter (in micrometers) and resolution parameter target_mpp define the scaled image tile size.

  • thumbnail_downsample_factor A factor used to reduce the WSI dimensions to create a low-resolution slide representation.

  • check_focus Run DeepFocus module to assess focus (blurryness) of the whole slide image.

  • deepfocus_model_path Path to DeepFocus checkpoint to use.

  • stain_normalization Whether to do any stain or color normalization.

  • stainnet Path to checkpoint for stain normalization model.

  • macenko_normalization If true, then use Macenko stain normalization. If false, use StainNet color normalization. This parameter is ignored if stain_normalization is false.

  • stain_reference_image Reference image (or a small patch, e.g., 2000 by 2000 pixels) to use with Macenko stain normalization.

  • stain_patch_size Macenco stain normalization patch size.

  • mask_background_cutoff Parameter for detecting image background with HoVer-Net.

  • pixel_mask_threshold_low Parameter for detecting tissue pixels on the low-resolution image.

  • pixel_mask_threshold_high Parameter for detecting tissue pixels on the low-resolution image.

  • fraction_for_mask Fraction of pixels in tissue required to call tile in tissue.

  • use_provided_grid Whether to use the grid provided in the input sample sheet. If false and no Space Ranger alignment is done, then a new grid of tiles is generated based on the grid parameters.

  • grid_type Type of the grid of tiles to generate. it can be hex, square, or random.

  • grid_spot_diamter Diameter of the spot (dimension of a tile) in micrometers.

  • grid_spot_horizontal_spacing Horizontal center-to-center distance between adjacent spots (or tiles).

  • grid_aspect_correction Factor to correct Visium slide aspect ratio.

  • overlap_scale_factor Imaging features extraction parameter. If the factor is 1, then features are extracted from the tile of the ST spot dimension.

  • hovernet_segmentation Do HoVer-Net segmetation. If false do StarDist segmentation.

  • nuclei_segmentation_dir name of directory to save segmentation.

  • hovernet_batch_size Parameter of HoVer-Net segmetation. This parameter is ignored when segmentation is done with StarDist.

  • hovernet_num_inference_workers Parameter of HoVer-Net segmetation. This parameter is ignored when segmentation is done with StarDist.

  • hovernet_chunk_size Parameter of HoVer-Net segmetation. This parameter is ignored when segmentation is done with StarDist.

  • hovernet_tile_size Parameter of HoVer-Net segmetation. This parameter is ignored when segmentation is done with StarDist.

  • stardist_model Path to checkpoint of stardist model.

  • stardist_block_size Size of the image block to run segmentation. Blocks are merged internally at the end of segmentation.

  • stardist_expand_size Size of cytoplasm arouhd nucleus in pixels.

  • hovernet_spot_assignment_factor Used for either HoVer-Net or StarDist segmentation postprocessing. Scaling factor of the boundary limiting the inclusion of nuclei to an ST spot. A value of 1 means the boundary size equals ST spot size.

  • hovernet_spot_assignment_shape Used for either HoVer-Net or StarDist segmentation postprocessing. The shape of the boundary, either square or disk.

  • hovernet_min_cell_type_prob Used for either HoVer-Net or StarDist segmentation postprocessing. This filtering parameteris used to remove nuclei assigned with low confidence.

  • extract_tile_features Extract (generate) imaging features for all tiles.

  • extract_inception_features If extract_tile_features then do Inception V3 features.

  • extract_transpath_features If extract_tile_features then do TransPath features.

  • extract_uni_features If extract_tile_features then do UNI features.

  • extract_conch_features If extract_tile_features then do CONCH features.

  • transpath_features_model One of 'CTransPath' or 'MoCoV3'.

  • use_conch_normalizer Use specialized CONCH normalizer, instead of the standard normalizer used with UNI and CTransPath.

  • uni_model_checkpoint Path to downloaded CONCH checkpoint. Download requires registration https://huggingface.co/MahmoodLab/UNI/blob/main/pytorch_model.bin.

  • conch_model_checkpoint Path to downloaded CONCH checkpoint. Download requires registration https://huggingface.co/MahmoodLab/CONCH/blob/main/pytorch_model.bin.

  • do_superpixels Do superpixel segmentation using SNIC algorithm.

  • export_superpixels_contours If true, export superpixel contours in JSON format.

  • superpixel_compactness Superpixel compactness parameter, see details of SNIC algorithm.

  • pixels_per_segment Number of pixels per superpixel segment, i.e., superpixel size.

  • superpixel_patch_size Superpixel patch size. Warning: patches boundaries are kept flat.

  • superpixel_downsampling_factor Superpixel downsampling factor for the input image downsampling .

  • od_block_size Block size for OD calculation.

  • expand_nuclei_distance Distance in pixels to expand the nuclei mask.

  • export_image Export the resized and normalized image in OME-TIFF format.

  • export_image_metadata Export input image metadata in OME-XML format.

  • compression Compression library to use with OME-TIFF, e.g., 'LZW'.

  • downsample_expanded_tile Downsample expanded tile.

  • expansion_factor Tile is read from expanded area around the tile center.

  • subtiling If true, split tile into subtiles, then extract features and compute average across the subtiles.

  • subcoords_factor Factor that defines the size of subtiles.

  • subcoords_list Centers of the subtiles within a tile.

  • do_clustering Do dimensionality reduction and clustering. Generate spatial and UMAP plots of imaging feature clusters as well as nucler morphometric features and classification results.

  • expansion_factor_for_clustering Features of the specified expansion factor are used for clustering.

  • suffix_for_clustering Features of this type are used for clustering.

  • plot_dpi DPI (dots per inch) of the figures.

  • hovernet_device_mode GPU or CPU device for use with HoVer-Net.

  • ctranspath_device_mode GPU or CPU device for use with TransPath inference models.

  • sample_tiles_subworkflow Run a subworkflow where a small number of tiles is saved, along with the HoVer-Net classification data.

  • tiles_per_slide Number of randomly selected tiles to use in the sampling tiles subworkflow.

  • do_segmentation_anndata DEPRECATED parameter, will be removed in future.