Table 4. Calculation of data characteristics in R.

Name of data characteristic	Name in matrix (data.prop) summarizing all data characteristics	Calculation in R	Dimension
dat.cpm Counts per million normalized and log transformed data		edgeR::cpm (dat, log=TRUE, prior.count = 1)	mxn
Feature sparsity	data.prop$P0_feature	apply (dat==0,1,sum)/ncol (dat)	m
Sample sparsity	data.prop$P0_sample	apply (dat==0,2,sum)/nrow (dat)	n
Feature mean abundance	data.prop$mean_log2cpm	apply (dat.cpm, 1,mean,na.rm=T)	m
Feature median abundance	data.prop$median_log2cpm	apply (dat.cpm,1, median,na.rm=T)	m
Feature variance	data.prop$var_log2cpm	apply (dat.cpm, 1, var)	m
Library size	data.prop$lib_size	colSums (dat)	n
Sample means	data.prop$sample_means	apply (dat,2,mean)	n
Sample correlation	data.prop$corr_sample	cor (dat, dat, method="spearman",use="na.or.complete")	nxn
Feature correlation	data.prop$corr_feature	cor(t (dat), t (dat), method="spearman",use="na.or.complete")	mxm

Data characteristics that are vectors or matrices. These are used in Table 3 to calculate the final 43 integer value characteristics. Here, dat is a (mxn)-Matrix, with features as rows and samples as columns.