Investigation of logicle transformation on CyTOF data

raw scale

chnls <- c("(Ir191)Dd", '(La139)Dd')
fr <- fr[,chnls]
densityplot(~., fr)

Estimated Logicle parameters

est_trans <- estimateLogicle(fr, channels = chnls)
#estimated parameters
sapply(chnls, function(chnl)
  sapply(c("t","m", "a", "w"), function(param)as.numeric(format(as.vector(environment(est_trans@transforms[[chnl]]@f)[[param]]), digits = 2)))
  )

##   (Ir191)Dd (La139)Dd
## t  24760.00  42855.00
## m      4.50      4.50
## a      0.00      0.00
## w      0.07     -0.05

About these paramters:

• t: Top of the scale data value,

• m: full width of the transformed display in asymptotic decades.

• a: Additional negative range to be included in the display in asymptotic decades.

• w: the linearization width in asymptotic decades.

Apply est_trans to the data

fr_trans <- transform(fr, est_trans)
densityplot(~., fr_trans)

Apparently something is wrong with flowCore on this logicle transform

range(fr_trans)

##        (Ir191)Dd   (La139)Dd
## min    -1.154779    -1.15396
## max 24760.000000 42855.00000

apply(exprs(fr_trans), 2, range)

##         (Ir191)Dd   (La139)Dd
## [1,]    -1.154779    -1.15396
## [2,] 24760.263672 42855.91016

Trying to fix it by flowCore::estimateLogicle + flowWorkspace::flowJoTrans

trans <- transformList(from = chnls,
                        lapply(chnls, function(chnl){
                              evn <- environment(est_trans@transforms[[chnl]]@f)
                              flowJoTrans(maxValue = evn[["t"]], pos = evn[["m"]], neg = as.numeric(evn[["w"]]), widthBasis = evn[["a"]])
                            })
                      ) 
fr_trans <- transform(fr, trans)
densityplot(~., fr_trans, channels = chnls)

Works!