estCutoffs {CATALYST} | R Documentation |
For each sample, estimates a cutoff parameter for the distance between positive and negative barcode populations.
estCutoffs(x, ...) ## S4 method for signature 'dbFrame' estCutoffs(x)
x |
a |
... |
optional arguments. |
For the estimation of cutoff parameters, we considered yields upon debarcoding as a function of the applied cutoffs. Commonly, this function will be characterized by an initial weak decline, where doublets are excluded, and subsequent rapid decline in yields to zero. In between, low numbers of counts with intermediate barcode separation give rise to a plateau. As an adequate cutoff estimate, we target the point that approximately marks the end of the plateau regime and the onset of yield decline. To facilitate robust cutoff estimation, we fit a linear and a three-parameter log-logistic function to the yields function:
f(x) = d / (1 + exp(b * (log(x) - log(e))))
The goodness of the linear fit relative to the log-logistic fit is weighed with:
w = RSS(log-logistic) / (RSS(log-logistic) + RSS(linear))
and the cutoffs for both functions are defined as:
c(linear) = - beta0 / (2 * beta1)
c(log-logistic) = argmin x { | f'(x) | / f(x) > 0.1 }
The final cutoff estimate is defined as the weighted mean between these estimates:
c = (1 - w) x c(log-logistic) + w x c(linear)
Will update the sep_cutoffs
slot of the input dbFrame
and return the latter.
Helena Lucia Crowell crowellh@student.ethz.ch
Finney, D.J. (1971). Probit Analsis. Journal of Pharmaceutical Sciences 60, 1432.
data(sample_ff, sample_key) # assign preliminary IDs re <- assignPrelim(x = sample_ff, y = sample_key) # estimate separation cutoffs re <- estCutoffs(x = re) # view exemplary estimate plotYields(re, "A1")