Oncofusion-NF pipeline

A pipeline for pathology slide analysis: tiles, detects tissue, infers tile labels, lymphocyte cell detection, and feature extraction.

Requirements

• Unix-like operating system
• Java 11

Quickstart

1. Install docker

2. Install nextflow:

   curl -s https://get.nextflow.io | bash
   

3. Launch the pipeline execution using docker

   ./nextflow run msk-mind/oncofusion-nf -profile standard,docker --samples_csv samples.csv
   

   samples.csv is a csv file where the first column is the slide ID and the second is the path to the slide.

4. When the execution completes, results will be in the results directory

Azure Batch Support

1. Create an Azure storage account and batch account.

2. Modify the nextflow configuration files as necessary (see https://www.nextflow.io/docs/edge/azure.html)

3. Set the necessary secrets using nextflow secrets set. At a minimum set AZURE_BATCH_KEY and AZURE_STORAGE_KEY.

4. Launch nextflow:

   nextflow -Dcom.amazonaws.sdk.disableCertChecking=true run main.nf -bucket-dir {azure_bucket_dir} -profile docker,cloud
   

   where {azure_bucket_dir} is an azure path like az://test/nftest.

5. When the execution completes, results will be in the results directory