SpaCST
SpaCET is a toolbox mainly focused on cancer areas in spatial transcriptomics.
First, SpaCET estimates malignant cell fractions based on a gene pattern dictionary of copy number alterations (CNA) and malignant transcriptome signatures across common tumor types.
Second, SpaCET deconvolves nonmalignant cell fractions and adjusts cell densities under a unified linear model. Using scRNA-seq datasets from diverse cancer types, we defined reference expression profiles of immune and stromal cells in a hierarchical lineage
Third, SpaCET infers intercellular interactions based on cell colocalization and ligand–receptor co-expression analysis.
What interest me most is cell interaction analysis.
For a spot, an L–R network score is defined as the sum of expression products between all L–R pairs, divided by the average random value from 1000 randomized networks. P values were calculated with the empirical null distribution generated from network scores of randomized L–Rinteractions.
Network Score \((\mathrm{NS})=\frac{\sum_i E_{L i} \times E_{R i}}{\left\langle\sum_i E_{L i} \times E_{R i}\right\rangle}, P\) value \(=P_r\left(\mathrm{NS}_{\text {random }} \geq \mathrm{NS}\right)\)
\(E_{Li}\) and \(E_{Ri}\) donate the expression of ligand and receptor from the ith L–R pair, respectively. The \(<>\) represents averaging the product sums from 1000 random networks.