Drop-seq public software distribution
Version 1.13
December 2017
dropseq@googlegroups.com


Included in this release of the Drop-seq toolkit is a new Java program
(CollapseBarcodesInPlace), as well as collection of R functions. This code has been
released to show examples of how we call cells using the cumulative distribution plot, and
plotting code we use to assess the quality of single-organism and barnyard-style
experiments. While we hope this R code is informative for your own projects, we cannot
directly support it.

CollapseBarcodesInPlace is a utility to collapse cell barcodes by edit distance.  This has
been a requested feature for the community, and we're happy to support answering questions
about how it works and how it's best utilized.  While we are confident that it performs
the correct task computationally, setting the parameters on what cell barcodes should be
collapsed in a way that yields additional UMIs per cell without over-collapsing cell
barcodes in barnyard experiments is unclear.  In our experiments, we have seen appreciable
yields in the number of reads gained by collapsing cell barcodes, but because the
collapsed cell barcodes share highly overlapping transcriptomes with the same UMIs, the
number of new UMIs gained is very low.

There are 4 different R functions included in this release (along with their associated helper functions):

selectCellByReadsSCM
categorizeCellsUsingKnee
plotSingleOrganism
plotPairOrganism

These have been used by our lab to select cells based on the cumulative read distribution
curve, and to plot various QC metrics in both single organism experiments
(plotSingleOrganism) as well as barnyard experiments (plotPairOrganism).
