RNA Detection

3.3.2 Subsetting and
Normalization of RNA-Seq
Data

1. Select only the genes that are represented in the atlas, and
   create aRNA_seq.csvfile, containing in rows the cells that
   you want to map against the atlas, and in columns, the genes
   for which expression data is present in the atlas.


2. For normalization, calculate tpm [11] for the final dataset.

3.3.3 Spatial Mapping:
Formatting and Loading
the Data

1. Load the downloaded analysis functions in R:

source("spatial_mapping.R")

2. The data should be organized in the following files:
   atlas: table containing binary expression data values forN
   voxels (rows) andMgenes (columns), such as one created in
   Subheading3.1.4.
   3D_coordinates_atlas: table containing the spatial coordinates
   for the voxels in the atlas. This table should contain no header.
   RNA_seq: table containing normalized RNA-seq counts values
   for the quality-filtered cells in rows and the genes in columns.
   This table should contain only the genes common between the
   RNA-seq profiles and the selected atlas, and the order of the
   genes must be identical in theRNA_seq.csvandatlas.csvfiles.


3. Load your data in R:

atlas<- read.table("atlas",header¼TRUE,sep¼"\t")

coordinates<- read.table("3D_coordinates_atlas",header¼FALSE,sep¼",")

rna_seq<- read.table("RNA_seq",header¼TRUE,sep¼"\t")

3.3.4 Spatial Mapping:
Specificity Index

1. In order to filter out noise and better match the RNAseq and
   ISH data, we define a specificity ratio (seeNote 7),rc,mfor each
   cell-gene combination:

rc,m¼

Dc,m

1

C

PC

a¼ 1 Da,m

whereC
Mis the read count matrixDwithCcells and M

genes, so thatDc,mdescribes the normalized number of reads

mapped to cellcfor genem.

The specificity index is calculated at the spatial mapping

script step:

specificity_matrix<- specificity_scores(rna_seq)

116 Kaia Achim et al.