README.md

Overview

To perform index hopping removal a rough guideline is provided here.

Preprocessing

First, cellranger needs to be run on all samples that were sequenced together as if the Snakemake workflow would run on these files. Note that ADT and GEX data needs to have seperate cellranger runs in order for this approach to work.

clean_and_filter_swapped.R

Needs the following input:

1. Path to the output Rdata file.
2. Path to a folder where all output files can be stored.
3. FDR threshold for cell calling (e.g. 0.01)
4. library size threshold for cell calling (e.g. 500)

ADT Data

Unfortunately, the output format of the above script is not a valid format to be processed with the gExcite pipeline. Therefore, the next two scripts have to be run to get appropriately formatted matrix, barcode and feature files. The order of the input files can be found in the scripts.

clean_adt_barcodes_indexHopping.R

Output is a clean barcode and a clean matrix file.

clean_adt_features_indexHopping.R

Output is a clean feature file.

GEX Data

Unfortunately, the ouput feature file of the above script contains Ensembl IDs instead of HGNC Gene Symbols. Using the following script helps to regenerate the appropriately formatted feature file.

map_geneSymbols_indexHopping.R

Unzip the features.tsv.gz, barcodes.tsv.gz, matrix.mtx.gz files. Run the R script "map_geneSymbols_indexHopping.R" to receive the correctly formatted feature.tsv file.

Postprocessing

After above files have been generated, the pipeline can be started by placing the corrected matrix.mtx, feature.tsv (with tag with_gene_names), and barcode.tsv files into the appropriate cellranger folder of the pooled samples. The files need to be renamed or provided via softlinks to make sure the name of the files is what Snakemake expects after parsing the sample map file.
Snakemake should then recognise these files as already generated and start with the next step, without trying to create an uncorrected cellranger output.