PBMC

Analysis automated pipeline

Clone the repository

git clone [email protected]:Melbourne-COVID-Predict/jascap.git

Enter in the jascap directory within R and activate renv, for reproducibility

module load R/4.2.1

open R in the terminal:

install.packages("renv")
 renv::activate()
 renv::restore()

Install makeflow

# Load miniconda
module load miniconda3

# run this command once
$ conda create -n cctools-env -y -c conda-forge --strict-channel-priority python ndcctools

# run this command every time you want to use cctools
$ conda activate cctools-env

Create input and result directories

# Define dataset directory, for example test_pipeline
master_directory=test_jacap
mkdir $master_directory
result_directory=$master_directory/results
reports_directory=$master_directory/results/reports
mkdir $result_directory
input_directory=$master_directory/input
mkdir $input_directory

# The R directory in the same location where you cloned jascap
code_directory=~/PostDoc/jascap

# This is the location of the metadata, which should match by `sample` with the column in the seurat objects. The metadata file needs to be in a different directory than the input file.
metadata_path=~/PostDoc/covid19pbmc/data/3_prime_batch_1/metadata.rds

# This is in a shared location 
reference_azimuth_path=/stornext/Bioinf/data/bioinf-data/Papenfuss_lab/projects/reference_azimuth.rds

Note: If no metadata.rds is available: create a data frame with 2 columns (sample | batch) with the samples matching the SAMPLE_NAME in the input file, >as described below. Set batch as Sample_name.

sample	batch
spleen	spleen
liver	liver

Add input files in the input directory.

1. Each input file should be a seurat object
2. Each input file should just include row counts in the RNA assay
3. Each input file should include only one sample
4. Each input file should be named SAMPLE_NAME.rds
5. All SAMPLE_NAME should be unique for each sample (if a sample has different time points the SAMPLE_NAME should be made unique, for example SAMPLE_NAME_TIME_POINT)

You should copy those files in your input directory, for example

cp myfiles/* $input_directory

Execute pipeline using makeflow

The available modalities are

• preprocessing (filtering and annotation without integration)
• fast (preprocessing + differential transcription, tissue composition, and cell communication analyses)
• complete (still not ready)

The available tissue annotation are

• pbmc (preprocessing based on Seurat Azimuth pbmc annotation)
• solid (preprocessing based on SingleR blueprint annotation)
• atypical (preprocessing based on unsupervised clustering)

# Create $input_directory/pipeline.makeflow
#
# The pipeline has 4 modes (first argument of create_pipeline_makefile.R)
# preprocessing, fast_pipeline, slow_pipeline, complete
Rscript $code_directory/R_scripts/create_pipeline_makefile.R complete pbmc unfiltered $result_directory $reports_directory $input_directory $code_directory $metadata_path $reference_azimuth_path

# Execute makeflow
conda activate cctools-env
makeflow -J 200 -T slurm $result_directory/pipeline.makeflow 

Monitor progress

makeflow_monitor $result_directory/pipeline.makeflow.makeflowlog