Workflow for generating predicted labels on a library of pre-trained garnett-cli classifiers. Run as a part of control workflow that generates predictions using a variety of tools.
This workflow relies on monocle-scripts to initialise a CDS object from input expression data. This CDS object is then used as an input to the classifiers. A list of prediction tables in standardised format is created as output.
To run the workflow, you will need to have nextflow installed. It is recommended to run the workflow in a clean conda environment. Specify the input parameters in nextflow.config.
Expression data are supplied as stadnard 10X-type directories that must contain three files: matrix.mtx, genes.tsv, and barcodes.tsv. Specify the path to this directory in the query_10x_dir parameter in config file.
Then run the following commands:
conda install nextflow
nextflow run main.nf -profile <profile>
The -profile flag can be set to 'local' or 'cluster' depending on where you're running the workflow.