Box 2. GAIN genotyping quality standards and procedures
From the following article
The GAIN Collaborative Research Group
Nature Genetics 39, 1045 - 1051 (2007) Published online: 29 August 2007
doi:10.1038/ng2127
- The Perlegen genotyping center will include one standard HapMap sample on half of the 96-well plates and one study sample duplicate (a different duplicate sample on each plate) on the other half of the plates, along with two parents of a study sample, when available, on each plate to make a trio for studies not based on trio samples
- The Broad Institute genotyping center will include one standard HapMap sample on each plate, along with one study sample duplicate (with a different duplicate sample on each plate) and two parents of a study sample.
- Samples with fewer than 90% of the SNPs called will be removed.
- At least 90% of the SNPs in a study will meet the following minimum data quality standards; actual data quality is expected to be much better:
- Hardy-Weinberg deviation P value > 0.00033 in any plate
- Call rate minimum = 90% and average
97% - For HapMap quality assessment samples, average call rates for both heterozygotes and homozygotes 97%
- Concordance in duplicate samples of 99.5%
- Quality scores meet a pre-determined minimum level, to be decided for each study by the Genotyping Group
- Minor allele frequencies meet a pre-determined minimum level, to be decided for each study by the Genotyping Group
