Class HapGraph

java.lang.Object
  alun.util.Main
      HapGraph

public class HapGraph
extends Main

This program estimates a graphical model for linkage disequilibrium or allelic associateion from genetic data. There are a number of ways to use it.

Usage : java HapGraph < haplotypes

where

• haplotypes is a file containing observed or reconstructed haplotypes. The file must contain a line for each haplotype with the alleles of each haplotype separated on the line by white space. The alleles must be specified as positive integers.
Used this way, with the data being read from the standard input file, the program will assume that the haplotypes are perfectly observed with no missing data.

Usage : java HapGraph haplotypes

where

• haplotypes is a file containing observed or reconstructed haplotypes. The file is as specified above except that 0 can be used to indicate a missing value.
Used this way, with the data being read from a file named on the command line, the program will iterate between inferring the missing data given the current graphical model for LD and re estimating the graphical model given the imputed data.

Usage : java HapGraph input.par input.ped [-ped] [-1]

where

• input.par is a LINKAGE parameter file
• input.ped is a LINKAGE pedigree file
Used in this way, HapGraph will estimate a graphical model for linkage disequilibrium from diploid data specifed in the linkage files. Again the program iterates between inferring complete, phase know, data given the current model for LD and re estimating the model given the currently imputed phase known data.

The default mode is to assume that the individuals in the pedigree file are an unrelated population sample regardless of any pedigree information in the file. This allows far faster computations to be made because each individual's complete phase known data can be updated as a single blocked Gibbs update.

If the -ped option is specified the individuals are placed in a pedigree as given in the input file. This requires that updating phases involves several separate blocked Gibbs updates on the pedigree, which is compuationally demanding.

If the -1 option is specified the first locus is ignored. This is because it is often a putative disease locus and an LD model may be wanted for marker loci only. For creating an LD model for inputting to McLink use

java HapGraph input.par input1.ped -1
to create an LD model in the file ld.model say. Then run McLink using
java McLink input.par input2.ped 1000 -v ld.model
or similar. input1.ped and input2.ped are a random population sample of genotyped individuals suitable for estimating an LD model from, and a pedigree on which to perform linkage analysis with LD.

The mouse, key and button controls are similar to those for ViewGraph with a few extra options.

• An additional slider bar controls the simulated annealing temperature.
• Additional text fields show the number of iterations, current score, best score, and time taken.
• There is an additional button to stop the search. Once stopped, further buttons appear to control output.
• There is a button to output reconstruted haploytpes. There is one haplotype per line. If haplotypes were input the output is in the same order. If diplotypes were input there are two reconstructed haplotypes for each diplotype in the same order as input.
• There is a button to output the estimated graphical model. There are other programs, such as HaploFreqs , that read this output file and provide summaries.

Constructor Summary

HapGraph()

Method Summary

static void	main(java.lang.String[] args)
static void	usage()

Methods inherited from class alun.util.Main

strip

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

HapGraph

public HapGraph()

Method Detail

usage

public static void usage()

main

public static void main(java.lang.String[] args)