Newport iServer MicroServer iTHX-M, Operator'S Manual

Page: 23 / 414

Chapter 1 | GTYPE Overview
15
100K AND 500K MAPPING ARRAYS
The Affymetrix GeneChip ® Human Mapping 100K and 500K array
sets provide an efficient method for genotyping more than 100,000
SNPs using a two-array set. Each SNP that can be detected with the
array has a group of probes, known as a probe set. The Dynamic Model
(DM) algorithm and the Bayesian Robust Linear With Mahalanobis
(BRLMM) algorithm are used to make genotype calls for the
GeneChip 100K and 500K Array data.
DM (Dynamic Model) Algorithm
The Dynamic Model (DM) mapping algorithm makes genotype calls
for a SNP. This algorithm is primarily being used to perform single-
array QC. For more information about the algorithm, see
Appendix C,
Dynamic Model Mapping Algorithm, on page 335 .
BRLMM (Bayesian Robust Linear With Mahalanobis) Algorithm
The Bayesian Robust Linear Model With Mahalanobis (BRLMM)
distance classifier algorithm implements a model based algorithm to
generate genotype calls from Affymetrix SNP microarrays. BRLMM is
a multi-sample, clustering method that fits probe effects across
multiple experiments. The following are BRLMM improvements over
DM:
• Increased precision in estimates of signal from each allele
• Genotype calls using a clustering approach
• Leveraged information from SNP calls across multiple samples
and across multiple SNPs to produce improved genotype calls
• Improved overall performance (call rates and accuracy)
• Equalized performance on homozygous and heterozygous
genotypes
Since BRLMM is a model-based algorithm, it needs to run on multiple
CEL files each time in order to estimate probe effects and SNP cluster
parameters. For mapping 500K data, it is advisable to run analysis on
at least 50, and ideally 100, biologically distinct samples.
GTYPE_manual.book Page 15 Wednesday, August 2, 2006 11:31 AM