Genetic Association Interaction Analysis (GAIA)

Statistical Interaction Analysis for Case-Control Data

Upload the first gene file
     

Upload the second gene file
     


 Input format for each input file is standard linkage format
 FAMID ID FATHERID MOTHERID SEX AFFECTIONSTATUS MK1_1 MK1_2 MK2_1 MK2_2...
There are sample input files that you can try out; GENE 1 TESTFILE IS HERE and GENE 2 TESTFILE IS HERE . To run these files through the analysis save these files to your machine (right click and select Save as... on most browsers) and upload them into the browse boxes above.

 Screening
GAIA can be used to screen your data so that only markers that are nominally significant (in single locus tests) are included. To do this load your data as normal (load your single file into both boxes, if necessary) and follow the instructions at the bottom of the next page.

 Notes on input file format
- If the input files contain different individuals then the program automoatically selects individuals who are present in both files. For compatibility of IDs across different Cardiff University schizophrenia databases the strings "SZC" and "SCZ1104" are automatically removed from the input files (so take care that your IDs do contain these strings as part of a particular individuals' unique ID). Note that input file 1 can be the same file as input file 2 if you have all the markers of interest in a single file. If the markers of interest are in the same file simply load the file in twice in the browse boxes
- Markers must be diallelic and coded 1 and 2.
- Missing genotypes must be coded as 0; if one allele is missing then the other alelle must also be missing.
- The marker names in the header line should not contain spaces or the underscore character "_". To identify the 2 alleles the program interprets anything after an underscore as the allele coding; e.g. somemarkername_1 somemarkername_2 describes alleles 1 and 2 at marker "somemarkername". Header line entries such as mkname_allele1 mkname_allele2 or mkname_all1 mkname_all2 etc should also work. The different alleles must be labelled differently!
- The individuals do not have to be included in the same order in the two files (however if the only difference between the files is the difference in order of individuals then the marker selection page will list all markers twice)
- Parents will usually be coded as unknown for case-control data; note the program simply ignores the values in the FATHERID and MOTHERID columns.
- Affected individuals must have a "2" in the AFFECTION_STATUS column and unaffected individuals must have a "1" in the AFFECTION_STATUS column.

If you use GAIA in a publication please cite the following paper
Macgregor S, Khan IA. GAIA: an easy-to-use web-based application for interaction analysis of case-control data. BMC Med Genet. 2006 Apr 5;7:34.
The full text (which gives more details on the methodology and implementation) is available from the above link.

Successful uses of GAIA include the following publications

The source code for GAIA is available here
Whilst efforts have been made to make GAIA work in the same way on different web browsers you may prefer to use the Mozilla/Firefox browser as Microsofts Internet Explorer browser (versions 6 and below) tends to continually ask you if you want to refresh the page...

Stuart Macgregor (contact: stuart.macgregor 'at' qimr.edu.au) and Imtiaz Khan, Biostatistics and Bioinformatics Unit, Cardiff University and Queensland Institute of Medical Research , 2005-2007