Phenocycler 2.0 pipeline

Prepares data and does 2D segmentation of Phenocycler images using deepcell method.

Usage example

cwltool pipeline.cwl subm.yaml

or to run it using SLURM on the HIVE using singularity:

sbatch -p GPU --gpus 1 --mem=500G cwltool --singularity pipeline.cwl subm.yaml

data_dir to be set in the subm.yaml to point to the directory containing the dataset

Requires a *metadata.tsv file containing the

Requires a *.channels.csv indicating which channel should be used for Nuclear and Cell segmentation according to the specification:

Field Name	Field Type	Recommend for Search (available on the portal) Backfill?	Definition	Options: required, optional	NOTES
channel id	text		Structure of the identifier depends on the acquisition system. Whenever possible this should exactly match the channel ID in the QPTIFF file. For example the channel ID in an QPTIFF might be something like "Channel:0:13" which would then be the value entered here.	required
is channel used for nuclei segmentation	boolean		Specifies if this channel can be used by algorithms performing nuclei segmentation ("Yes" or "No").	required	In data types for which a HIVE pipeline performs cell and nuclear segmentation, exactly one channel must have value "Yes".
is channel used for cell segmentation	boolean		Specifies if this channel can be used by algorithms performing cell segmentation ("Yes" or "No").	required	In data types for which a HIVE pipeline performs cell and nuclear segmentation, exactly one channel must have value "Yes".
is antibody	boolean		Does this channel encode image intensities associated with an antibody ("Yes" or "No")? If "Yes" then the relevant antibody is expected to be listed in the antibodies.tsv file with the same channel ID.	required

Channel:0:0,Yes,No,No Channel:0:1,No,No,Yes Channel:0:2,No,No,Yes Channel:0:3,No,No,Yes Channel:0:4,No,Yes,Yes Channel:0:5,No,No,Yes Channel:0:6,No,No,Yes Channel:0:7,No,No,Yes Channel:0:8,No,No,Yes Channel:0:9,No,No,Yes Channel:0:10,No,No,Yes Channel:0:11,No,No,Yes Channel:0:12,No,No,Yes Channel:0:13,No,No,Yes Channel:0:14,No,No,Yes Channel:0:15,No,No,Yes Channel:0:16,No,No,Yes

The expected input directory structure:

data_dir/
├── experiment.qptiff
├── experiment.channels.csv
├── experiment.metadata.tsv
│ 
└── extras
      ├── antibodies.tsv
      └── contributors.tsv

The output structure:

pipeline_output/
├── expr
│   └──  experiment_expr.ome.tiff
└── mask
    └──  converted_mask.ome.tiff