/infercnv-snakemake

Primary LanguageR

A simple Snakemake pipeline to run inferCNV on SingleCellExperiment objects.

Step 1: Install relevant Python packages

pipenv install --python 3.8
pipenv shell

Step 2: edit config file

input_rds: peng-scRNASeq-manually-filtered-sce-tumour-normal-assigned.rds
annotation_column: cell_type
sample_column: donor

tumour_type: Tumour epithelial
normal_type: Normal epithelial

cutoff: 0.1
denoise: TRUE

samples:
  - CRR034499
  - CRR034500

Key entries:

  • input_rds: input SingleCellExperiment containing both tumour and normal. Gene names should either be [SYMBOL] or [ENSEMBLID]-[SYMBOL]
  • annotation_column: what column of colData(sce) says which cells are tumour vs normal?
  • sample_column: what column of colData(sce) refers to sample (patient/donor)? The pipeline runs inferCNV once per donor
  • tumour_type The ID in colData(sce)[[annotation_column]] that refers to the cells that are tumour/malignant
  • normal_type The ID in colData(sce)[[annotation_column]] that refers to the cells that are 'normal'
  • cutoff,denoise parameters passed to inferCNV
  • samples list of samples (patients/donors/etc) to run inferCNV over. Should be present in colData(sce)[[sample_column]]

Step 3: Run the pipeline

snakemake -j1 --configfile config/peng-test.yml --use-singularity --singularity-args "--bind /home/campbell/share/:/home/campbell/share/"

replacing /home/campbell/share with your directory.

Step 4: Inspect results

Notes

  • This pipeline aggregates genes to gene symbol by summing