By Monica Heger
A sequencing-based assay that assesses copy number could become a low-cost tool to look for cancer biomarkers and assess drug effectiveness, according to researchers now at the Center for Translational Oncology and Immunology in Mainz, Germany.
In a study published in BMC Genomics, the researchers, who at the time they did the research were at Merck's Rosetta Inpharmatics group, developed an assay to assess copy number variation using a low-coverage sequencing strategy of nuclear, telomeric, and mitochondrial DNA.
John Castle, lead author of the study and now the head of genomics and bioinformatics at the Center for Translational Oncology and Immunology, said the method could cost as little as $100 per assay and promises a low-cost way to diagnose diseases such as mental retardation, where copy number variation is a key characteristic. By comparison, some array-based tests to analyze copy number cost from several hundred dollars to over $1,000.
Some researchers argue, however, that the strategy of trying to reduce cost by sequencing to lower-fold coverage will quickly become obsolete as sequencing costs drop.
The researchers created sequence libraries from genomic DNA of healthy males and females, and genomic DNA from a lung carcinoma cell line from a male. The samples were barcoded so they could be multiplexed, and each genome was sequenced to about .01-fold coverage. They generated about 3.2 million reads from the female pool, 6.7 million reads from the male pool, and 5.2 million reads from the lung carcinoma pool, all with read lengths of 33 bases. In each pool 92 percent to 93 percent of the reads aligned to the genome.
To assess copy number, they first divided the genome up into blocks. Then, by comparing the number of reads in a given block to the number of reads in the corresponding block from the reference genome, they were able to estimate copy number. For instance, if a block was 100 kilobases wide and the reference sample had 100 reads, but the experimental sample had 300 reads, then the copy number would be six for the diploid genome, Castle said.
The resolution is dependent on the number of reads, he said. Five million reads will yield a 70-kilobase resolution. "So if you want more resolution, you sequence more reads," he said. "It's a very flexible, tunable method."
The technique is also more sensitive than using microarrays, Castle said. Particularly in tumors, where copy number can be very variable, sequencing is more accurate because microarrays lose their accuracy with higher copy numbers.
"In tumors, you can get copy numbers up to 100, and beyond a copy number of six, microarrays don't reflect the right quantity," he said.
Castle's team also used the method to assess mitochondrial and telomeric copy number. He said those measurements could be useful for looking at the effect of drugs. For instance, he said, researchers could use the method to see if cancer drugs are changing mitochondrial DNA expression levels.
Looking at telomeric copy number could help gain a better understanding of cancer etiology. In their study, the researchers identified copy number reductions in the telomeres of the lung cancer cell line. "Somehow cancer cells and cell lines have figured out a way to not die when their telomeres go to zero," Castle said.
In terms of diagnostics, the method could be useful for disorders such as mental retardation and other conditions characterized by aneuploidy in certain regions of the genome, Castle said.
Mark Gerstein, a professor of biomedical informatics at Yale University who is involved with the 1,000 Genomes Project, said that while there are a number of different assays for assessing nuclear copy number, the fact that this one also assesses mitochondrial and telomeric copy number sets it apart.
"Normally those bits of the genome are neglected," he said. Figuring out mitochondrial copy number could be important for determining functional changes, he added.
Rebecca Leary, a graduate student at the Johns Hopkins Kimmel Cancer Center who has used sequencing to identify rearrangements that act as biomarkers to determine tumor drug response (IS 2/23/2010), said that while the method of evaluating copy number would be useful, she wasn't sure that this particular method would have clinical implications.
"Reducing sequencing costs by creating lower coverage libraries may have been appropriate earlier, but the new technologies and improvements will make that less of an obstacle in the near future," she said. Also, the resolutions that the researchers achieved — between 60 and 100 kilobases — "are lower than many of the chips available that assess copy number."
Yale's Gerstein said that while the price of sequencing is dropping, it is still cost prohibitive. "Determining copy number from low coverage sequencing is important now because it's expensive to sequence things at high coverage."
Gerstein and Leary did agree that copy number would be an increasingly important component to measure when studying disease and when looking for disease and drug response biomarkers. Compared to gene expression, said Leary, "copy number is more absolute. Gene expression studies are still highly variable."
Also, added Gerstein, researchers are beginning to think that structural variation plays a larger role in disease than other types of variation. There has been an increased interest in assessing copy number, particularly as researchers begin to recognize that structural variation may be "the major component of human genomic variation, possibly more important that SNPs," he said.