⚠️ This repo has been archived. Please use nf-cellbender

Cellbender

Repo containing scripts for running cellbender

Directory structure should be as followed:

top level directories - actions, data, work

actions - contains all the scripts which are used to run souporcell

data - contains subdirectories for each sample with sampleid as name, output is written to these subdirectories

work - contains logs subdirectory which stores all the log files for running cellbender

Getting Data

Ensure the file irods.txt contains 2 tab separated fields. The first contains a sample IDs and the second contains a path to the location on irods. Each row contains a single sample ID and its corresponding irods path as shown below:

HCA_A_LNG11599017_and_HCA_A_LNG11986504	/seq/illumina/cellranger-arc/cellranger-arc201_count_308b45a587ebfa6f443e3bb754bb625f
HCA_A_LNG12177503_and_HCA_A_LNG11986505	/seq/illumina/cellranger-arc/cellranger-arc201_count_aa07d3102ccae90a449dca434e47b104

Run the script get-cellbender.sh inside the data directory and it will download all the relevant data. If you have different file names to the default edit get-cellbender.sh as the comments instruct. I recommend using a screen session for this process. To run the script simply:

cd data
../actions/get-cellbender.sh

Running cellbender

Ensure your sample ID list is contained within the samples.txt file. It should be a single field file containing one sample ID per row. I.e.

HCA_A_LNG11599017_and_HCA_A_LNG11986504
HCA_A_LNG12177503_and_HCA_A_LNG11986505

Run the script bsub-cellbender.sh inside the work directory.

cd work
../actions/bsub-cellbender.sh

If you need to change the resources used then edit the bsub-cellbender.sh script as appropriate. If you want to change the cellbender parameters then edit the cellbender-0-2-0.sh script.

Jobs will be submitted to the FARM, they can be monitored with the command bjobs or you can look at the log and error files within the logs subdirectory.

Quality Control

Running the script cellbender.qc.R will assess the quality of the output. It runs on a single cellbender output directory at a time. We do not have singularity image that can be used to run the scripr right now, so here is temporary solution:

export R_LIBS_USER=/nfs/cellgeni/pasham/R/%p-library/%v
cd work
/software/cellgeni/wrappers/r-4.3.1/Rscript cellbender.qc.R -v -m 3 . 

Multiomes

If running cellbender on multiome samples. Run the bsub < bsub.extract.gex.from.multiome.sh script and then use the cellbender-matrix_multiome.sh to run cellbender.

If mtx input was used

Cellbender produces h5 files that cannot be read by scanpy if it used mtx as input (STARsolo or multiomes). So, run bsub < actions/cellbender/scripts/bsub.cellbender_output_to_mtx.sh to transform cellbender output into mtx. It will find all cellbender outputs in working directory and adds mtx folder to them.