Evaluation for estimated cell type proportions

Source: R/supportingfunctions.R

Several evaluation metrics are provided, such as mean absolute deviance (`MAD`), Kendall-tau correlation coefficient (`Ken`), Pearson correlation coefficient (`Cor`), given true cell type proportions.

evaluate(est.prop,true.prop)

Arguments

est.prop

The estimated cell type proportions.

true.prop

The True cell type proportions

Value

Cell-type level evaluations based on MAD, Ken, and Pearson (`cell.type.eva`), and overall evaluations based on averaged MAD, Ken, and Pearson (`all.eva`).

Examples

##read data
library(InteRD)
readRDSFromWeb<-function(ref) {readRDS(gzcon(url(ref)))}
urlremote<-"https://github.com/chencxxy28/Data/raw/main/data_InteRD/"
pseudo.seger<-readRDSFromWeb(paste0(urlremote,"pseudo.seger.rds"))
true_p<-readRDSFromWeb(paste0(urlremote,"true_p.rds"))
SCDC_ENSEMBLE_MAD<-readRDSFromWeb(paste0(urlremote,"SCDC_ENSEMBLE_MAD_seger.rds"))
evaluate(SCDC_ENSEMBLE_MAD,true_p)$all.eva
#>     all.mad   all.ken   all.cor
#> 1 0.1069572 0.7358974 0.8939433