ConstructMIMOSAExpressionSet: A wrapper for constructing an Expression Set for MIMOSA
In RGLab/MIMOSA: Mixture Models for Single-Cell Assays

Calls a series of other functions that will reshape and refactor the data frame into the right format for use by MIMOSA Standardized for use with internal SCHARP data sets. We provide some default arguments as examples. Currently slow, and very much prototype code.

ConstructMIMOSAExpressionSet(
  thisdata,
  reference = quote(STAGE %in% "CTRL" & PROTEIN %in% "Media+cells"),
  measure.columns = c("Neg", "Pos"),
  other.annotations = setdiff(colnames(thisdata), measure.columns),
  default.cast.formula = component ~ ...,
  .variables = quote(.(PTID, TESTDT, ASSAYID, PLATEID)),
  featureCols = 1,
  ref.append.replace = "_NEG"
)

`thisdata`	is the input data frame
`reference`	is an `expression` that evaluates to a `logical` vector which specifices the observations in the data frame that are to be used for the negative control or reference set
`measure.columns`	is a `chracter` vector that specifies which columns hold the observed counts
`other.annotations`	is a `character` vector that specifies which additional columns in the data frame should be included in the returned data. By default we take everything, but you could specify only relevant phenotypic information.
`default.cast.formula`	is a `formula` that tells reshape how to recast the data frame so that rows corresponde to different measured components and columns correspond to samples. By default `component~...` will put the components as the rows (i.e. positive and negative cell counts) and all measured phenotypic information on the columns.
`.variables`	is a dotted list that specifies the variable names (columns of the data frame) by which to group the data when organzing stimulated and unstimulated observations. i.e. PTID x ANTIGEN x TCELLSUBSET x TESTDT, or something else for your own data.
`featureCols`	is a `numeric` vector that specifies the indices of the columns to be used to name the features. If the casting formula is `component~...` then there is only one feature column (and it is the first one), so `featureCols = 1`, by default.
`ref.append.replace`	the terminating character string in the column names of the negative controls. It will be replaces with _REF for 'reference'

data(ICS)
E<-ConstructMIMOSAExpressionSet(ICS,
  reference=ANTIGEN%in%'negctrl',measure.columns=c('CYTNUM','NSUB'),
  other.annotations=c('CYTOKINE','TCELLSUBSET','ANTIGEN','UID'),
  default.cast.formula=component~UID+ANTIGEN+CYTOKINE+TCELLSUBSET,
  .variables=.(TCELLSUBSET,CYTOKINE,UID),
  featureCols=1,ref.append.replace='_REF')