Medland, SE, Neale MC (submitted) An integrated phenomic approach to multivariate allelic association.
Example Usage:
./WriteMx_U.pl -p unrelated.ped -d example.dat
./WriteMx_F.pl -p fullsibs.ped -d example.dat
WriteMx_U.pl WriteMx_F.pl
Example Mx scripts
Current version 1.1 Last updated June 2010 Bug fixes


WriteMx is a perl program designed to facilitate the integrated phenomic approach to multivariate allelic association. The script takes are input Merlin format .ped and .dat files and outputs Mx format data and script files for running the analysis in Mx (which can be freely downloaded for a range of operating systems from the Mx homepage

There are two versions of this program:
WriteMx_U.pl is designed for use with data from unrelated individuals
WriteMx_F.pl is designed for use with data from family samples composed of full-siblings and/or pairs of twins. This program is currently designed to handle families with up to five siblings.

Input files are expected to contain more than one trait - technically there is no maximum number of traits, however analyzing a large number of traits in the analysis will increase the analysis time and may lead to optimization difficulties. Analyses work best with 5 or less traits.

Ped files can contain more than one snp. The Mx scripts produced contain loops to allow the analysis of multiple snps. Note the run time increases as a function of the number of snps and the current version of the program has not been tested with more than 100 snps.

The program also provides summary information on the number of snps, traits and covariates entered into the analysis, sample size, sex ratio, and in the case WriteMx_F information regarding family structure.

The program is designed to work with either continuous or binary phenotypes (coded as affection or 1/0) but will not take a combination of these 2 data types as Mx is currently unable to analyse mixed continuous/ordinal models.

The Mx scripts constain some generic starting values. However, it is a good idea to customise these for your own data, especially if you are working with binary data. You may also encounter optimisation problems if your data contains only a few families of a given sibship size, i.e. 200 pairs and 4 trios. In this situation you may wish to consider analysing the pairs and trios within the same group.

Last Update 6 Nov 2008 by Sarah Medland - please contact me (sarahMe [at] qimr.edu.au) if you find errors. Please note all photos appearing on the webpage headers were taken by me - please ask before using them for something else.
Design inspired by Nautica05 writen by studio7designs.com for OSWD