Virtual Slicing in 3D Data

While majority of the current spatial omics technilogies are 2D in nature, real 3D technologies are emerging. In 2018, Wang et al. created a 3D spatial expression dataset consisting of 28 genes from 32,845 single cells in a visual cortex volume using the STARmap technology.

To better understand 3D data and enable smooth swtich between 3D and 2D views, we created a set of functions in Giotto to generate virtual 2D slices (cross section) in 3D data.

Usage

Main steps

The main steps are:

• Create neighborhood network
• Choose a method of defining a 2D plane and identify the corresponding parameters
• Call CrossSection routine to generate the slices and visualize them

1. Spatial network

To define spatial relationship between cells, we first create a spatial network using createSpatialNetwork function.

createSpatialNetwork function

STAR_test <- createSpatialNetwork(gobject = STAR_test, dimensions = 'all', method='Delaunay',delaunay_method = 'delaunayn_geometry')

Here the spatial network is created using the Delaunay triangulation method. For 3D data, the option dimensions='all' will select all 3 dimensions to create the spatial network. The option delaunay_method = 'delaunayn_geometry', calls the "delaunay" function from the "geometry" package to do the 3D triangulation.

2. Define a 2D plane with the right parameters

Giotto offers four different ways of defining the location and orientation of a 2D slice, i.e. "equation", "3 points", "point and norm vector" and "point and two plane vectors".

createCrossSection function

createCrossSection <- function(gobject,name="cross_section",spatial_network_name = "Delaunay_network", thickness_unit = c("cell","natural"), slice_thickness = 2,cell_distance_estimate_method = "mean", extend_ratio = 0.2, method=c("equation","3 points","point and norm vector","point and two plane vectors"), equation=NULL, point1=NULL,point2=NULL,point3=NULL, normVector=NULL, planeVector1=NULL,planeVector2=NULL, mesh_grid_n = 20, return_gobject = TRUE)

The method parameter

• method="equation" choice:

Requires the parameter equation to be set. For example, equation=c(0,1,0,600). This command will create a 2D slice specified by the equation 0 * X + 1 * Y + 0 * Z = 600 (or simply Y = 600). In general, if the "equation" parameter is set to be c("A","B","C","D"), the plane defined by Ax+By+Cz=D will be created. This is the default option and is also the most flexible one to use.

• method="3 points" choice:

Requires the paramters point1, point2, point3 to be set. For example: point1 = c(0,600,0), point2 = c(1,600,0), point3 = c(0,600,1). This command will create a 2D slice uniquely specified by three points "point1", "point2" and "point3". This option will be helpful if the equation of the slice plane is not easy to determine but the coordinates of a few anchor points for the slice in the tissue sample is available (or can be easily estimated).

• method="point and norm vector" choice:

Requires the parameter point1 and normVector to be set. For example: point1 = c(0,600,0), normVector = c(0,1,0). This command will create a 2D slice uniquely specified by the point "point1" and a vector that is perpendicular (or "norm") to the plane. This option will be helpful if the orientation (direction) of the slice plane is predefined and one anchor point is available.

• methd="point and two plane vectors" choice:

Requires the parameters point1, planeVector1 and planeVector2 to be set. For example: point1 = c(0,600,0), planeVector1 = c(1,0,0), planeVector2 = c(0,0,1). This command will create a 2D slice uniquely specified by the point "point1" and two vectors that are parallel to the slice plane and not parallel with each other. This option will be helpful if the required parameters for the other options are not available.

Other important parameters for defining a virtual slice

3. Call CrossSection routine

Here we walk through an example. First the preparation steps.

# from STARmap dataset
## create consistent color code
mynames = unique(pDataDT(STAR_test)$cell_types)
mycolorcode = Giotto:::getDistinctColors(n = length(mynames))
names(mycolorcode) = mynames
## creat spatial network 
STAR_test <- createSpatialNetwork(gobject = STAR_test)

Next create the cross section.

## creat virtual slice
STAR_test = createCrossSection(STAR_test,method="equation",equation=c(0,1,0,600),extend_ratio = 0.6)

Then create the cross section plots.

## view the location and orientation of virtual slice in 3D (cell type)
insertCrossSectionSpatPlot3D(STAR_test, cell_color = 'cell_types', axis_scale = 'cube', point_size = 2, cell_color_code = mycolorcode)
## view the location and orientation of virtual slice in 3D (Gene expression)
insertCrossSectionGenePlot3D(STAR_test, expression_values = 'scaled',axis_scale = "cube", genes = "Slc17a7")
## view the cells captured by virtual slice (3D, cell type)
crossSectionPlot3D(STAR_test,point_size = 2, cell_color = "cell_types",  cell_color_code = mycolorcode,axis_scale = "cube")
## view the cells captured by virtual slice (2D, cell type)
crossSectionPlot(STAR_test, point_size = 2, cell_color = "cell_types",cell_color_code = mycolorcode)
## view the cells captured by virtual slice (3D, gene expression)
crossSectionGenePlot3D(STAR_test, point_size = 2,genes = c("Slc17a7"), expression_values = 'scaled')
## view the cells captured by virtual slice (2D, gene expression)
crossSectionGenePlot(STAR_test, genes = "Slc17a7",point_size = 2,cow_n_col = 1.5, expression_values = 'scaled')