camiolomj/ICLite: Immune Cell Linkage Through Exploratory matrices

ICLite performs deconvolution of bulk sequencing data using the composition of a mixed cell population. It requires log ratio transformed compositional data from cellular assays such as mass cytometry or other high-dimensional flow cytometry technologies to identify a best fit solution for a data set. Genes are clustered together using the blockcluster package after a multistep transformation process that includes construction of correlation matrix and removal of transcripts based on cutoffs for number and strength of interactions. These parameters, as well as number of clusters assumed for the blockcluster package, are varied by the user to generate numerous solutions which are then evaluated using a weighted scoring system. The optimal fit solution is identified based on maximization of Integrated Completed Likelihood (ICL) from gene clustering and connectivity between gene modules and cell populations. Package outputs include CSV files of the gene modules and correlation plots of gene module scores versus cell counts.

install.packages("remotes")
remotes::install_github("camiolomj/ICLite")