By Monica Heger
Merck, GlaxoSmithKline, and Johnson & Johnson were among several pharmaceutical companies to present preliminary results from proof-of-principle sequencing studies at last week's Next Generation Sequencing and Genomic Medicine Applications Summit in Burlingame, Calif.
Researchers from the companies expressed interest in using the technology to aid in drug development, the search for disease biomarkers, and in selecting patients for clinical trials.
Instead of focusing on whole-genome sequencing, which they all said was still too expensive, the three pharma firms are performing targeted sequencing studies — focusing on either specific genes, a portion of the exome, or the transcriptomes.
Johnson & Johnson: Assessing Capture Techniques
Johnson & Johnson has been testing capture and sequencing platforms and is interested in doing multiplexed targeted resequencing studies of oncogenes for mutation and copy number variation analysis. Peter Verhasselt, principal scientist in translational genetics and genomics at Janssen Pharmaceuticals, a Belgium based subsidiary of Johnson & Johnson, presented preliminary results from a study comparing different capture and amplification techniques with the Roche 454 and Illumina sequencing platforms.
Verhasselt said the company sequenced 558 targets in 44 different genes, covering about 90 kilobase pairs from 24 different cancer cell lines from the National Cancer Institute's panel of cancer cell lines. The specific regions comprise genes frequently mutated in cancer, such as tumor suppressors and oncogenes, as well as druggable targets like kinases. He said the goal is to use sequencing to look for new drug targets and biomarkers.
"The prices are too expensive to do whole-genome sequencing on large cohorts, especially if you are only interested in a subset of the genome," Verhasselt said in a presentation at last week's meeting. But, with targeted sequencing, you can barcode and index, enabling samples to be pooled, he said.
In a proof-of-principle study, his group tested three different amplification and capture techniques — the RainDance RainStorm PCR technology, Agilent's SureSelect in-solution technique, and method from a Belgian company called Multiplicon — with sequencing on both the 454 and Illumina Genome Analyzer.
"The technologies are still new," Verhasselt said. "So we did a proof of principle to see if we can find mutations and copy number variations in cell lines [using the different technologies], and if so, how well they perform and how costly they are."
The team aimed to capture 95 percent of the targets at 40-fold coverage. Verhasselt found that each combination of capture and sequencing techniques had its own advantages and drawbacks. He said that of the capture technologies, Agilent's SureSelect gave the best coverage, while Multiplicon had the most noise.
The samples enriched by RainDance and Multiplicon were shotgun sequenced with Illumina using 36-base reads, while SureSelect-enriched samples were sequenced with 76-base paired-end reads. RainDance- and Multiplicon-enriched samples were also sequenced with 454 technology, but Verhasselt said there were "technical problems" in combining Agilent's SureSelect with 454, so the team did not have data from that combination.
The known mutations were verified in all combinations, although sequencing with Illumina's shorter 36-base reads failed to detect a 5-base deletion because the reads were too short to map the deletion accurately. "But, patterns are still consistent across all approaches," Verhasselt said.
While the company has both an Illumina GA and a 454 GS FLX, Verhasselt said for this project the company has decided to stick with Illumina sequencing because of its higher throughput and lower cost. Verhasselt said the team has not yet made a decision on which capture technique to use, and is working with the different providers to improve the specificity and uniformity of the capture.
Verhasselt said that the study demonstrated that the technology works, and that the company is now deciding on the specific questions it wants to answer. The goal is to be able to do targeted resequencing from biopsy samples instead of cancer cell lines. He said the group did not have plans to do whole-genome sequencing, but was interested in RNA-seq, methyl-seq, and in the future single-cell sequencing.
[pagebreak]
GlaxoSmithKline: Variants for Drug Targets
John Whittaker, director of statistical genetics at GlaxoSmithKline's Essex, UK-based facilities, presented early results from an effort to sequence 200 genes in 15,000 people for a range of diseases, including cardiovascular, respiratory, psychological, immune, and neurodegenerative diseases.
Whittaker said that the goal is to use sequencing to find variants in targets for which the company already has drugs, and also to use the variants to select people for clinical trials. In the old drug-development method, researchers start with a trait or disease, search for the causative gene, and then develop a drug, Whittaker said. "But, given that it takes 10 to 20 years to bring a drug to market, we don't have time to do this path," he said. "Now, we go in reverse."
As a result, instead of starting with a trait or disease, and finding the causative gene and a drug that targets it, the GSK researchers are starting with a drug, and looking for variants within that target. "It's an ideal question to answer with sequencing," Whittaker said.
The GSK team has so far sequenced about 200 genes in 15,000 people using a paired-end sequencing strategy on the Illumina GA with 75 base pair reads, to an average 30-fold coverage. They used NimbleGen's capture array and BGI's SOAP aligner and SOAPsnp for variant calling.
So far, the researchers have only analyzed results from about 5,000 patients and have found about 50,000 SNPs, 7,000 of which are unique and non-synonymous.
It's still too early to tell if any novel associations have been found, said Whittaker. "There's nothing that's completely convincing yet, but we see a lot more small associations than we would expect to see by chance."
Merck: RNA-seq for Schizophrenia
Merck, meantime, is focusing its sequencing efforts solely on transcriptome sequencing.
"When we think about designing a drug that works at the protein level, we think there's a lot of value in studying RNA," Tomas Babak, a senior research scientist at Merck, said at the conference. "Not that we don't think studying DNA is important, but at the end of the day, unless your loci is in an obvious genomic location, you will spend a lot more work to get to the protein."
Babak said his team is working on a proof-of-principle study using RNA-seq to measure allele-specific expression in samples from schizophrenia patients. Babak said that the team starts with total RNA, instead of poly-A-plus RNA, and uses engineered hexamers to remove ribosomal RNA. The sequencing is done on the Illumina GA, and the approach gives them strand specificity.
The researchers tested the technique on 20 human samples of schizophrenia — 14 cases and 6 controls. While there were too few samples for the findings to be statistically significant, Babak said it was a "very informative exercise," and that the team plans to "apply the approach to thousands of datasets as they become available."
In the study, the team found 170,000 variants, 75 percent of which are in dbSNP. They found variants in pathways associated with neurodegenerative disease, cytoskeleton remodeling, and neurofilament development.
Babak said that the team expects to have data from additional RNA-seq studies of schizophrenia by the end of the year. The company has one Illumina GA, and Babak said that the sequencing would likely all be done in house, but that some of it could be outsourced.