Stefan Schrod; Niklas Lück; Robert Lohmayer; Stefan Solbrig; Dennis Völkl; Tina Wipfler; Katherine H. Shutta; Marouen Ben Guebila; Andreas Schäfer; Tim Beißbarth; Helena U. Zacharias; Peter J. Oefner; John Quackenbush; Michael Altenbuchinger

Table 2.

Estimated spatial partial correlations versus the association strength determined by SpaCeNet for the four top-ranked associations discussed in the main text

	$\text{[math]}$	$\text{[math]}$	$\text{[math]}$	Δρ⁽¹⁾
Flt1–Mbp	−0.560	−0.557	−0.558	0.327
Ctgf–Gja1	−0.118	−0.119	−0.119	0.102
Ctgf–Pcp4	−0.227	−0.234	−0.231	0.083
Reln–Sst	−0.174	−0.205	−0.190	0.078

Here, $\text{[math]}$ corresponds to spatial partial correlations between variable j and $\text{[math]}$ conditioning on $\text{[math]}$ and k. An analogous definition holds for $\text{[math]}$ and the respective average is given by $\text{[math]}$ . SpaCeNet's leading order association strength is given by Δρ⁽¹⁾ with positive signs corresponding to negative associations in line with the definition of the precision matrix of ordinary GGMs.

Spatial Cellular Networks from omics data with SpaCeNet

This Article

Preprint Server

Current Issue

In This Issue