|
Not apply any filters to the CGH array data sets. However, given the
high density of probes per chromosome on the CGH arrays, visual
interpretation of known genomic lesions in plots of raw data are unclear
by even a small percentage of outlier probes. Hence, we applied a 50-kb
moving average, as calculated below, to plots presented in Figs. 4–6.
Smoothing: The log2 ratio measured for all
m
probes of the chromosome by using the following weighted moving average.
 
yi is the measured
log2 ratio at xi (formula 1).
Triangular
function (formula 2).
xi
is the chromosomal position (in bases) of probe i.
W
is the half width of
the triangular function.
To explore error
modeling of aCGH data, the colorcoded points in CGH array plots by using
the following method. The variance of each smoothed point is estimated
as follows:
σ
is the standard deviation of the log2 ratio for each set of
chromosome-specific probes on the CGH array in a series of
hybridizations with cell lines that are diploid for the entire
chromosome under investigation.
In each example, probes with log2 ratios >
are
expressed in red (increase in copy number), probes with log2 ratios <
are expressed in green (decrease in copy number), and no change probes
are shown in blue.
To use the
expression array to measure copy-number changes in four well
characterized tumor cell lines (COLO 320DM, HT 29, MDA-MB-231, and
MDA-MB-453) with chromosomal abnormalities. The lesions catch in the
tumor cells include a high-level (log2 ratio = 6.4)
amplification of MYC in COLO 320DM and an amplicon
spanning 8q23.1–24.23 with a 3-fold (log2 ratio = 1.5)
increase in the copy number of MYC with simultaneous single-copy
8p deletion in HT 29 (Fig. 1).
To test
oligonucleotide aCGH with in vivo
material, we used the expression arrays to screen four soft-tissue
sarcomas analyzed prior with cDNA aCGH. A compare method using a common
build of the human genome to map probe positions demonstrated a striking
similarity between the cDNA and oaCGH profiles (Fig. 2).
The reference
46,XX DNA which the initial measurement of the performance of arrays was
done with hybridizations of genomic DNA samples from a series of cell
lines with variable copy numbers of the X chromosome.
The comparison of
the median ratios for the X-chromosome probes from these hybridizations
on the expression array (373 probes) and the CGH array (4,878 probes)
showed that the correlation coefficients were 0.98 and 0.99,
respectively, and the slope increased from 0.47 to 0.96 (Fig.
3).
The median ratio values
on aCGH of the X-chromosome probes in these data were 0.5 for XY_XX, 1.0
for XX_XX, 1.4 for XXX_XX, 2.1 for XXXX_XX, and 2.6 for XXXXX_XX (Fig.
3 C and D).
These data are in the closest agreement with the ideal values of 0.5,
1.0, 1.5, 2.0, and 2.5 reported for aCGH. These data highlight with the
usefulness of designing and selecting in situ synthesized
oligonucleotide probes for aCGH assays.
Examined the
3,293 probes on these arrays representing unique genomic sequences at an
average spacing of 23 kb along chromosome 18 in the 18q-syndrome
patient-derived cell line GM50122 to assess our ability to detect and
map intrachromosomal single-copy losses. The single-copy loss on 18q
was finded, and the breakpoint region was localized visually and
numerically to the known genomic location by using a moving average of
50 kb and noise model for chromosome 18 (Fig. 4).
To detect the
single-copy losses, tumor cells may contain homozygous deletions that
target specific genes. Using the HCT116 colon carcinoma cells as a model
system to assess our ability to detect and map homozygous deletions by
using the aCGH. These cells include two known homozygous deletions on
chromosome 16 that include the A2BP1 gene (16p13.2) and the
FRA16D locus (16q23.2).
First,
of these homozygous deletions was detected by BAC aCGH.
Second,
homozygous deletion (16q) was characterized by positional cloning. Only
loci with probes on these arrays which measured log2 ratios
were below a theoretical threshold of -2 in replicate experiments were
16p13.2 (nine probes) and 16q23.2 (two probes) (Fig. 5).
|
