sage-bionetworks/ampad_lineage

Ordering patients on smooth manifolds using lineage estimation tools

ampad_lineage's Issues

trying to understand the BinLims assignment in get var genes

Hi all
thank you for the interesting pipeline, its a neat work that impresses me everytime :)
i was wondering if you could explain the rationale behind your get var genes function Get.High.Var.Genes, especially the part on binLims BinLims <- seq(log(10/length(M)),max(M),Bins + 1) ; i dont quite understand the rationale for setting start as the number of genes in the assay, but setting the greatest highest rowmeans max(M expression as the end for seq(). conventionally, i only noticed BinLims setting up using min and max(M) of expression values instead of the length of number of genes.

from your Misc.Processing script:

Get.High.Var.Genes <- function(Dat, Bins = 10, frac = .2){
  
  M = rowMeans(Dat)
  V = RowVar(Dat)
  
  BinLims <- seq(log(10/length(M)),max(M),Bins + 1)
  
  In = c()
  
  for (i in 1:Bins){
    In_red <- which(as.logical((M >= BinLims[i])*
                                 (M < BinLims[i+1])))
    temp <- V[In_red]/M[In_red]
    temp2 <- sort(temp, decreasing = T, index.return = T)
    I <- temp2$ix
    topX <- round(length(In_red)*frac)
    In <- c(In,In_red[I[1:topX]])
    
  }
  
  In <- In[!is.na(In)]
  In <- unique(In)
  return(In)
  
}