This R package is created and maintained by Nikos Karaiskos. DistMap can be used to spatially map single cell RNA sequencing data by using an existing reference database of in situs.

DistMap accompanies the following publication

The Drosophila Embryo at Single Cell Transcriptome Resolution,
Science 358, 194 (2017)

N. Karaiskos^#, P. Wahle^#, J. Alles, A. Boltengagen, S. Ayoub, C. Kocks, N. Rajewsky^& and R. Zinzen^&

^# Contributed equally
^& Corresponding authors: N. Rajewsky, R. Zinzen

Contact the author in case you've found a bug.

The easiest way to install DistMap is through devtools

library(devtools)
install_github("rajewsky-lab/DistMap")

Pay extra attention if you read data with different commands, for instance read.table against fread from the data.table package. Gene names which contain -s or other special characters might be read differently. In that case you'll probably encounter an error during binarization of the data (see below).

The DistMap object is used to store the following structures:

• raw.data, the raw data (e.g. UMI counts) of the experiment, provided by the user as a matrix with genes as rows and cells as columns.
• data is the normalized data, provided by the user as a matrix similar to the raw data.
• binarized.data is the binarized version of the single cell data computed via the binarizeSingleCellData function.
• insitu.matrix is the matrix of the reference database, provided by the user, with genes as columns and positions (bins) as rows. See the included example used in the paper.
• geometry, a matrix containing the cartesian coordinates of each bin in three dimensional space. Provided by the user, bins as rows and coordinates as columns, see geometry.txt. provided as an example.

The first step is to initialize the DistMap object

dm = new("DistMap",
         raw.data=raw.data,
         data=normalized.data,
         insitu.matrix=insitu.matrix,
         geometry=geometry)

Then the binarized single cell data is computed and the cells are mapped onto the reference atlas

dm <- binarizeSingleCellData(dm, seq(0.15, 0.5, 0.01))
dm <- mapCells(dm)

Once the cells have been mapped, the DistMap functions can be used to compute a vISH or a gradient of a gene and visualize the expression pattern

computeVISH(dm, 'sna', threshold=0.75)
computeGeneGradient(dm, 'sna')