FEEDBACK: getWeightMatrix()

[HenrikBengtsson]

Feedback / questions about:

getWeightMatrix(prop.max, prob.min, nb.arch, nb.samp, sparse.coeff = 0.5, contam.coeff = 0.6, contam.max = 30)

#' Generate weight matrix
#'
#' @param prop.max An integer that is the maximim proportion for a present archetype
#' @param prob.min An integer that is the minimum proportion  for a present archetype
#' @param nb.arch  An interger that is the number of archetypes (number of columns)
#' @param nb.samp An interger that is the number of samples (number of rows)
#' @param contam.max An interger between 0 and 100 that controls the maximal level of contamination by normal cells
#' @param contam.coeff A numeric between 0 and 1 that control the contamination level
#' @param sparse.coeff A numeric between 0 and 1 that control the sparsity by rows
#' @return A matrix of weights (the sum of rows is equal to 1)

• TYPO: prop.max vs prob.min
• What's the range of the above two arguments? (0,1,2,...,100)?
• Why integers and not fraction in [0,1]? Same question for contam.max.
• Several spelling errors.
• A bit random spacing.

[pneuvial]

@mpierrejean can you fix this? In particular, rename prop.max to prob.max and make sure that prop.max, prop.min, contam.max are in [0,1], not [0,100]? This would be consistent will the standard in acnr.

[pneuvial]

Also, it is unclear from the documentation what the difference between contam.max and contam.coeff is. Please clarify in the doc that the contamination is drawn from a binomial distribution with parameters size=contam.max (or 100*contam.max after fixing this issue) and prob=contam.coeff.

[HenrikBengtsson]

Is this issue moot? I don't see a getWeightMatrix() in intercept branch.

[HenrikBengtsson]

Flagging as 'expired'.