Biomedical Engineering Reference
In-Depth Information
r
An experiment is defined as the results of one Illumina array and the genotyping and LOH
analysis of all the DNA samples typed on the array. Every experiment folder contains 11
subfolders containing data for different steps in the analysis flow (see Figure 3.3a).
s
The final Genotyping report is generated in Beadstudio and contains genotyping and quality
information. The Locus-By-DNA report is exported from Gencall and contains information about
the SNPs, the linkage panel and the array used for the experiment.
t
Genomic information about all the SNPs is provided by Illumina in the oligonucleotide pool
assay manifest, a small CD provided in the Sentrix Array box.
u
LOH detection can be impaired by a low tumor percentage; normal cells and (tumor infiltrating)
leukocytes that are present in the isolate will result in diminished LOH calling. Ideally, pure
tumor material should be used for LOH detection, although in practice tumors will be tested
with up to 40 % of non-tumor cells. In these cases, most SNPs will still be called heterozygous;
however, their genotype quality will be reduced. We therefore use the quality scores that are
automatically assigned to each genotype by the gene calling program Beadstudio. Each SNP
gets two different quality scores; the sample-specific gene call score (GCS) and the SNP-specific
GTS. Since reduced quality scores in tumor tissue might indicate LOH, we calculate the GCS/GTS
ratio for each SNP and sample. For normal samples, we expect the ratio to be high (between
0.8 and 1). Therefore, only the high-quality heterozygous SNPs in the normal sample, with
a GCS/GTS ratio between 0.8 and 1.0, are included in LOH analysis. LOH is called for these
SNPs when the tumor genotype call is homozygous, or is heterozygous and has a GCS/GTS ratio
below 0.8.
v
For a heterozygous SNP with a GCS/GTS ratio
<
0.8 the SNP genotyping quality is defined as
'
low
.'
3.3 Troubleshooting
•
Frequently asked questions:
Illumina has generated a list of 'general problematic issues'
and suggestions on how to resolve the problems. This information can be found on
www.illumina.com under faqs (frequently asked questions) for DNA analysis.
•
Experimental design:
A typical experiment is the whole genome-genotyping and LOH
analysis of 12 tumors and paired normal DNAs. These 24 samples are genotyped using
four pools of approximately 1500 SNPs, the linkage panels. The SNPs are hybridized to
Sentrix arrays, a matrix of 96 arrays, each detecting 1500 SNPs. Thus, for each sample,
four arrays of approximately 1500 SNPs each are hybridized and subsequently combined
to approximately 6000 genotypes. These together form the linkage panel. Alternatively, a
genome-wide cancer SNP panel, or custom SNP panels of 1536, 768 or 384 SNPs, could
be tested.
•
Genotyping of FFPE tissue DNA using the Illumina Goldengate assay and Sentrix
arrays:
For each array, the Goldengate protocol requires 250 ng of activated (biotinylated)
DNA in the single-use DNA activation (final volume 10
μ
l) or 2
μ
g of DNA in the
multiple-use DNA activation (final volume of 100
l). Fan
et al
. [46] describe that less
DNA can be used. For the use of FFPE DNA, we routinely use 1
μ
g of DNA as input in
the multiple-use DNA activation and dissolve the resulting biotinylated DNA in 60
μ
μ
lof
RS1 buffer.
