PHILADELPHIA — The Association of Biomolecular Resource Facilities’ Microarray Research Group has completed a project that tests the ability of five different array platforms to detect genetic changes in a human leukemia sample using comparative genomic hybridization, and will soon submit the results of the study for publication and release the data to the public, according to a MARG member.
Laura Reid, director of research and development at Expression Analysis and part of MARG’s annual research project, discussed data from the study at IBC Life Sciences’ Discovery 2 Diagnostics meeting here this week. During her presentation, Reid showed initial data from five platforms that analyzed a common sample.
MARG’s data demonstrates that all five platforms were able to successfully detect the eight known genetic changes in HL-60, a human leukemia sample that has been previously characterized using karyotyping. A manuscript detailing MARG’s CGH work has been completed, and Reid said that she expects it to be published by the end of this year, perhaps in ABRF’s Journal of Biomolecular Techniques.
While the concept behind MARG’s CGH project — using the same sample to test the response of five rival array platforms — resembles the design of the first phase of the US Food and Drug Administration’s Microarray Quality Control project, Reid said that MARG’s work is in no way a comparison of the five platforms, but an attempt to provide core facilities with greater confidence that the platforms they are using in CGH are indeed working.
The MAQC project is a collaboration between government, academia, and industry that formed several years ago to test concordance between eight different array platforms. Results of the project culminated in a special issue of Nature Biotechnology that was published in 2006 (see BAN 9/12/2006). The MAQC results also fed several marketing campaigns by participants like Illumina, Affymetrix, and GE Healthcare who used data from the project to trumpet their platforms.
“I can see that it looks like an MAQC study because we are doing the same samples on multiple platforms,” Reid told BioArray News in an interview following her talk. However, she added, “it is not MAQC because it is not being hosted by FDA, and it’s not involving the manufacturers themselves.”
It’s also not similar to MAQC because the objective of MARG’s project was not to test concordance, and by extension performance, but to test the overall efficacy of microarray technology in characterizing a sample that had already been studied using older cytogenetics methods, said Reid.
Reid said that MARG decided to carry out the CGH study because of the rise of DNA-based array applications, like CGH or SNP genotyping, as opposed to the RNA-based applications that have dominated the market in the past.
“We had done RNA-based microarray research for a few years in a row and you could see that the winds were changing and it was time to look a little into the DNA-based assays,” she said. “When we started this project it was the end of 2006 and rumors were in the air about the 1-million marker arrays that were coming out from Affy and Illumina. So it seemed like a really good time to work up to that.”
The project was designed to test the ability of several widely used platforms to detect the same changes that had previously been seen in HL-60. The group selected five chips: a BAC array from the Roswell Park Cancer Institute in Buffalo, NY; an Agilent Human Genome CGH 44K Array; an Illumina HumanHap550 BeadChip; the Affymetrix GeneChip Human Mapping 500K Array Set; and the Affy Human Genome U133 Plus 2.0 Array.
HL-60 was selected as the sample because “it is a very convenient sample to use in the analysis and since it’s been karyotyped before, we could compare our results now with what previous labs have found,” said Reid.
“We had done RNA-based microarray research for a few years in a row and you could see that the winds were changing and it was time to look a little into the DNA-based assays.”
RPCI carried out testing using its BAC platform, while the Genomics Core Lab at the Memorial Sloan-Kettering Cancer Institute in New York tested the sample using the HumanHap550, Agilent 44K, and Affymetrix 500K sets. Meantime, MARG member Herbert Auer, who was director of the microarray unit at the Ohio State University Comprehensive Cancer Center at the time, tested the sample using the U133 array, which is commonly used in gene expression studies rather than CGH.
All platforms detected the copy number changes that had been previously characterized in HL-60, including the gene expression platform. Reid said that this was an encouraging result, considering that different software tools were used in the analysis portion of the experiment, and that the protocols for the experiments were highly varied, considering the nature of some of the platforms.
In terms of data-analysis tools, MARG used a circular-binary segmentation algorithm for the BAC platform and the U133 data, the Partek Genomics Suite for the Affy 500K data, and the manufacturers’ software for the Agilent and Illumina platforms.
Reid pointed out that this variation in protocol and analysis would make it difficult to compare the platforms on a performance basis. “This is what is so hard with the project in general,” she said. “The platforms are so different, the analyses are so different. Can you really compare apples versus oranges?”
She said that the data from the study will be available through the ABRF website as well as some public databases for those that would like to “dig down into the details.” However, rather than use the data to compare platforms, Reid said that her hope is that core facility directors will use the results in order to gain confidence in the results of array-based CGH experiments.
“Before I came to Expression Analysis I ran an academic core [at the University of North Carolina] and I know what it’s like when you have to make decisions between platforms and you only have a limited budget and you are not sure what one to choose and if you’ll get reliable data,” said Reid.
“I am hoping that at a minimum the MARG study will show you that you can at least get these known changes and that you can at least get repeatable results on the known platforms so that provides a level of confidence for the core facility directors,” she added.