[BioC] A friendly mention of LPE package

pingzhao Hu phu at sickkids.ca
Wed Feb 27 19:41:47 CET 2008


Dear Users,

If you use LPE package, this information may be helpful for you.

Based on my experience, LPE package is very useful to identify 
differentially expressed genes for data sets with very small
sample size.  One thing that did not/not clearly document in the user 
manual is that when
the sample size in each group is larger than 5, the results will keep 
changing when you run the SAME code many times.

I happen to find this comes from subrountine (am.trans) called by lpe function.

 > am.trans
function (y)
{
     if (ncol(y) > 5)
         y <- y[, sample(1:ncol(y), 5)]                  # here makes 
the results keep changing
     n <- ncol(y)
     if (n < 2) {
         stop("There are no replicated arrays!")
     }
     A <- c()
     M <- c()
     cc <- permute(1:n)
     for (i in 1:(n - 1)) {
         A <- c(A, c((y + y[, cc[i, ]])/2), recursive = TRUE)
         M <- c(M, c(y - y[, cc[i, ]]), recursive = TRUE)
     }
     return(cbind(A, M))
}

Therefore, when you use lpe pacakge and when one of your groups in 
your dataset has more than 5 samples, you shold notice
that your results will be changing. You can fix this problem by 
commenting out the code
    # if (ncol(y) > 5)
      #   y <- y[, sample(1:ncol(y), 5)]


Pingzhao








========================================
Pingzhao Hu
Statistical Analysis Facility
The Centre for Applied Genomics (TCAG)
The Hospital for Sick Children Research Institute
MaRS Centre - East Tower
101 College Street, Room 15-705
Toronto, Ontario, M5G 1L7, Canada
Tel.: (416) 813-7654 x6016
Email: phu at sickkids.ca
Web: http://www.tcag.ca/statisticalAnalysis.html



More information about the Bioconductor mailing list