Assessing Array CGH Platform Performance between Commercial Products
Assessing Array CGH Platform Performance between Commercial Products
A review of Resolving the resolution of array CGH.
Note: This is a review of the published article listed below. All information, quotes, figures, methods, and findings mentioned in this review are from that article, and are the property of its authors and/or the publication in which the article originally appeared.
A recent collaboration between researchers in British Columbia and The Netherlands (2007) compared array CGH performance between 10 different platforms, including the Agilent 244K microarray. The group characterized the various platforms through the introduction of "functional resolution" (a new metric that incorporates the distribution of array data points) and described a Java-based application "ResCalc," which automated the calculation of performance metrics for any aCGH platform (including any species and arrays covering only specific chromosomal segments). Results conclusively demonstrate that the Agilent oligo aCGH microarray exhibited the highest sensitivity of the oligonucleotide platforms, with a single element being sufficient to detect a single-copy alteration. In addition, the Agilent platform demonstrated the highest performance for single-copy alterations between 1 and 49 kb (8.7 to 97.5%), as well as compatibility with FFPE samples. Platform comparisons such as this study enable researchers to gain valuable insights into which platform is best suited to their individual research needs based on published results.
Fig. 1. Theoretical detection sensitivity.
(A) Detection sensitivity for each array platform was calculated based on the percentage of possible alterations of a given size that interact with at least one array element (blue bars). To determine the proportion of alterations of size nbp detectable by an array platform we first defined the set of all possible alterations (possible alterations are represented by red and green bars) of size n bp for all genomic regions covered by the array (excluding centromeres and acrocentric regions). We then calculated the percentage of alterations not detectable as those that are completely contained within each coverage gap. (B) Detection sensitivities for each platform are plotted for alteration sizes from 1 to 500 kb; the alteration size at which a platform exceeds a 95% detection rate defines the optimal sensitivity of that platform.
Fig. 2. Single-copy detection sensitivity.
(A) The BT474 cell line contains an average of 4.5 copies of each chromosome. Previous FISH studies characterized chromosome 8 into segments with 4, 5, and 6 copies. Comparing the ratios observed for 6 and 4 copies we can simulate the performance of a 3:2 copy number ratio. (B) Comparison of copy number profiles of chromosome 8 across three platforms. (C) Determination of the number of elements that must be pooled to allow detection of single-copy alterations. BT474 profiles were used to determine the number of elements that must be pooled to separate the average ratios for 4 and 6 copies by at least 1 standard deviation (indicated by *) for the SMRT, Agilent, VUMC, and Affymetrix platforms (the noise of the Affymetrix Mapping 10K is projected to be equivalent to the 100K and 500K sets due to the use of genomic reduction steps and identical oligonucleotide design strategy). (D) Single-copy alteration detection sensitivities for each platform are plotted for alteration sizes from 1 to 500 kb; the alteration size at which a platform exceeds the 95% detection rate defines the optimal sensitivity of that platform. Each platform was penalized based on the number of elements that must be pooled according to the calculation described in (C). Data were not available for the Nimblegen platform; the data are adjusted to incorporate pooling of five elements as described in Selzer et al. Similarly, data were not available for the Illumina platform; due to the similar probe length and labeling technology compared to the Affymetrix platform, a three-clone requirement was assumed.
Fig. 3. Breakpoint precision.
(A) The precision with which a breakpoint can be defined is derived from the genomic distance between each element end (as alteration boundaries can be defined to reside within an array element). The set of all interelement end gaps in the genome can then be determined (a to l) and sorted by increasing size. The percentage of the genome covered by interelement end gaps less than n bp in width (example size of gap "d") defines the proportion of breakpoints that demonstrate a precision of at least n bp (assuming 1 possible breakpoint per base pair). (B) Breakpoint precisions for each platform are plotted for alteration sizes ranging from 1 kb to 1 Mb; the precision level at which a platform exceeds 95% defines the optimal breakpoint precision of that platform.
Title: Resolving the resolution of array CGH.
Authors: Coe BP, Ylstra B, Carvalho B, Meijer GA, Macaulay C, Lam WL.
Journal: Genomics. 2007 Feb 1; [Epub ahead of print]
More