Oligonucleotide Arrays

Although the discussion in this chapter has concerned genomic clone arrays, some groups have used oligonucleotide arrays for analysis of copy number changes in tumors (33,34) and large-scale copy number variation in normal individuals (16). The signal to noise ratio for

3 I litrtion chr2(M5,720 ,?34--t 5.3ÎO.Î 33 - UCSC Genome Drowser v7i Mcioafi Internet [upbrer LjlGj I

3 I litrtion chr2(M5,720 ,?34--t 5.3ÎO.Î 33 - UCSC Genome Drowser v7i Mcioafi Internet [upbrer LjlGj I

Fig. 5. High-density fosmid clones. Example of the coverage of a small region of chromosome 20 with fosmid clones as shown in the UCSC genome browser.

genomic hybridizations to these arrays is generally low producing a large level of measurement variation such that direct hybridization of complete genomic DNA requires considerable data smoothing (e.g., running averages of eight data points) and statistical calling of copy number trends (34). However, reducing the complexity of the hybridization, typically by using restriction digestion with catch linker PCR, which favors only the smaller fragments, copy number information can be generated. Affymetrix arrays designed for SNP analysis (33) as well as custom long-oligonucleotide arrays (ROMA technology [35]) have been used in this way. Data generated from such arrays is still inherently noisy as differences in the restriction pattern of the test and reference DNAs produce artifactual differences, which might be interpreted as copy number changes. To overcome this, averaging or more sophisticated statistical analysis is used to identify copy number trends. In practice, three or more oligonucleotides within a genomic region are required to report a copy number change in order for this to be accepted as valid (16,33). Thus, although an array comprising 100,000 oligonucleotides would appear to have an average resolution of 30 kb, in practice the resolution is three times this, i.e., 90 kb. The advantage of oligonucleotide arrays is that each sequence can be designed to be unique and repeat free. Inevitably for tiling path arrays using large insert clones, many clones will include segmental duplications or high levels of repeat sequence which compromise the level and specificity of response of such clones to copy number changes.

Fig. 6. Array-comparative genomic hybridization (aCGH) analysis using different resolution arrays. aCGH profile of a microdeletion on chromosome 9 analyzed using a 1 Mb array (large gray bars) and with a custom high density fosmid array (small black bars) targeted to the deletion breakpoints.

Fig. 7. Array-comparative genomic hybridization (aCGH) analysis using different resolution arrays. aCGH profiles of the proximal breakpoint of the chromosome 9 microdeletion shown in Fig. 6, using the 1 Mb array (large black bar), the custom high density fosmid array (medium gray bars) and a high density small insert clones array (short black bars).

Fig. 7. Array-comparative genomic hybridization (aCGH) analysis using different resolution arrays. aCGH profiles of the proximal breakpoint of the chromosome 9 microdeletion shown in Fig. 6, using the 1 Mb array (large black bar), the custom high density fosmid array (medium gray bars) and a high density small insert clones array (short black bars).

Was this article helpful?

0 0

Post a comment