New version not limited to 2 individuals

Preamble

require(DNAmixtures)
data(USCaucasian,SGMplusDyes)
USCaucasian2<-rbind(USCaucasian,data.frame(marker="D18S51",allele=9,frequency=0.001659950))
ff<-c(paste(Sys.getenv('HOME'),'/Research/Perturb/Sep2013',sep=''),
    paste(Sys.getenv('HOME'),'/Peter-Julia/mixtures/UAF',sep=''))
for(f in ff) if(file.exists(f)) {setwd(f); break}
source('DNAmixtureUAF3ormore.R')

real.df<-read.csv("real.txt",header=T,fill=TRUE,sep="")
source('real.db.txt')

mixD<-DNAmixtureUAF(list(real.df),k=...,K=...C=list(...),database=real.db,dyes=list(SGMplusDyes),use.order=FALSE)

This differs from previous version in the following ways:

  1. new version of DNAmixtureUAF function (different filename, but called in the same way)
  2. must use use.order=FALSE option when calling DNAmixtureUAF

And the job above:

  1. is set up to use the data set Julia provided on 27 January (in file real.txt)
  2. uses specially created mini-database of frequencies (obtained by source('real.db.txt'))

Files: DNAmixtureUAF3ormore.R | real.txt | real.db.txt | real.hkb

Performance

The new code is not very carefully tested, but I am pretty sure it is correct. The network it creates follows our Beta-binomial algebra closely in notation. It seems to make a bigger network for the same problem than our earlier special-case coding for the 1 or 2 individuals case - see real.hkb using Hugin to see what I mean - and it seems to take a long time to run.

Suggested experiment(s)

Start with repeating a small example using 2 or fewer individuals to see that we get the same answers.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License