RNA Detection

This scoring system compares the binary vector of expression

provided by the whole mount ISH data with a binarized version of

the expression pattern for each sequenced cell. To binarize the

expression vectors, we used a threshold of 10 reads above which a

gene was considered expressed. This was configured by the variable

count_thresholdin the script above.

Mathematically, the scoreSec,erefbetween the binary expression

vectorecfrom single cellcandereffrom voxel ref in the ISH dataset

is defined as:

sc,ref¼

XM

m¼ 1

frc,m ec,m;eref,m



with

frc,m ec,m;eref,m



¼

trc,m



, ec,m¼eref,m¼ 1

trc,m



, ec,m¼1, eref,m¼ 0

0, Otherwise:

8

<

:

and

trc,m



¼

rc,m

1 þrc,m

This scoring scheme is designed to assess the correspondence

between a single cell and each reference voxel with regard to the

specificity ratio of each gene for the considered single cell. The

specificity scores are transformed to fall in the interval [0,1] follow-

ing an algebraic function,t, which avoids giving too much weight

to exceptionally specific genes and quickly reduces the weight of

nonspecific genes that may hinder the precision of the mapping.

3.3.6 Spatial Mapping:
Selecting the Confidence
Thresholds

1. For a single cellc, once the scores against every voxel in the
   reference dataset are computed and sorted, we need to define a
   score threshold above which we consider the voxels as the
   potential area where the single cell was located.
   To find this threshold, we will perform a simulation study
   by generating random “simulated single cells.” We start with
   100x coverage (100 simulated samples per sequenced cell).
   Each simulated single cell is created by randomly shuffling the
   specificity scores for all genes in each sequenced cell.
   The simulated dataset is generated at the spatial mapping
   script step:

generate_simulated_data(specificity_matrix,100,"simulated_data/")

2. The command above will create C datasets containing 100
   simulated cells each. Each dataset has two files:
   n.data: table of specificity scores
   n_bin.data: table of binary expression inferred from the spec-
   ificity scores

118 Kaia Achim et al.