There are two main components to this repo: the supplementary report for the paper Detection and Removal of Barcode Swapping in Single-Cell RNA-seq data (bcswap_supp.html) and the code used to generate it (all other files).
Unless you want to reuse any of the code or tweak/regenerate the report for yourself, you probably only want only the .html file of the report.
The preprint of the paper can be found here
You can also find a .pdf of a poster I have used to present this work at Single Cell Genomics 2017 (poster.pdf)
On my machine, Github will not show the html report interactively via the webpage: to view this document, it's easiest to download the entire repo as a zipped file (click "Clone or download" above, in green) or otherwise clone the repo using git itself.
To generate the .html file, you first need to download the data. This can be done by changing your working directory to the one that contains this repo,
then running the get_data.sh file. Then you can compile the .Rmd file either using RStudio
or rmarkdown::render(), from the R package rmarkdown. You'll need to ensure you have the packages listed in the first code chunk installed. This step takes around 20 minutes on a 2016 MacBook Pro.
If you are interested in the barcode swapping phenomenon, there are several other sources to read:
-
James Hadfield's blog post is, to our knowledge, the earliest identification of barcode swapping (December 2016). He also provides sensible experimental steps to reduce the severity of the swapping. He later published an update with more information.
-
The Sinha et al. paper on bioRxiv uses single-cell RNA-seq protocols to identify the source of the swapping.
-
Another bioRxiv paper, by Owens et al., did not identify any increased swapping on HiSeq X, using genome sequencing data from unbalanced heterozygotes.
-
Illumina released a white paper summarising the issue.
-
Larsson et al. describe a method to deconvolve the effects of swapping from plate-based scRNA-seq assays, and suggest a method for estimating swapping rates using the distinctive "crosshair" patterns.
-
van der Valk et al. do not identify excessive barcode swapping on HiSeq X (they also have an excellent and in-depth introduction to the barcode swapping phenomenon)
-
Costello et al. have produced a very comprehensive overview of the swapping they have observed in their lab for a variety of bulk sequencing studies.