Skip to main content
Premium Trial:

Request an Annual Quote

At Cold Spring Harbor Meeting, Baylor HGSC Reports on Use of Ion Torrent Proton, PGM

Premium

Baylor College of Medicine's Human Genome Sequencing Center has been testing Life Technologies' Ion Torrent Proton platform extensively, completing more than 360 runs so far and comparing it with Illumina's technology for exome sequencing. It has also developed a variety of applications for the Ion Torrent PGM, which has entered its data production line.

Earlier this month at the Personal Genomes and Medical Genomics conference at Cold Spring Harbor Laboratory, Donna Muzny, the HGSC's director of operations, detailed the center's experience with both platforms.

Baylor's HGSC currently has 16 Illumina HiSeqs, two Illumina MiSeqs, four Ion Torrent Protons, four Ion Torrent PGMs, three Roche/454 instruments, one PacBio RS, and a number of Sanger sequencers.

A beta version of the Proton arrived at the center in April and was upgraded to the commercial version in September. As of mid-November, the center had completed more than 360 runs on the Proton, testing the platform initially on a range of standard microbes, mammalian BACs, and libraries from human HapMap samples and primates.

Last month, it started testing 200 base pair reads on the Proton using improved beads that are expected to become commercially available in December. These beads offer better conjugation so they support longer libraries and attach more template. They also improve the data quality, resulting in a threefold increase of aligned Q20 bases, Muzny reported. As a result, the center now achieves between 9 and 10 gigabases of aligned data per run, including a recent 11.5-gigabase run, she said. The platform now produces aligned data in less than 22 hours, she said, including the four-hour run time.

More recently, the researchers have been evaluating the Proton for exome sequencing, and have compared results for the same sample to the Illumina HiSeq platform.

For that comparison, they sequenced the exome of a previously sequenced individual with Charcot-Marie-Tooth syndrome using Baylor's NimbleGen-based VCRome 2.1 exome capture design, the Proton's PI chip, and 200 base pair reads. The run yielded 8.7 gigabases of aligned data, an average coverage of 91x, and 92 percent of the target bases were sequenced with at least 20x coverage.

Sequencing the same sample on the Illumina HiSeq, using data generated within the past year in Baylor's Whole Genome Laboratory, which provides clinical sequencing, yielded 11 gigabases of data, or an average coverage of 170x. Of the target bases, 97 percent were covered with at least 20x coverage by the Illumina data.

Results from both platforms overlapped substantially: 94 percent of high-quality single-nucleotide variants were called in both datasets, though the lab has not tested the overlap for indel calls yet. Importantly, the Proton identified the same two pathogenic mutations that Baylor scientists previously reported in a publication in the New England Journal of Medicine.

Muzny said the center's goal for 2013 is to generate clinical-grade exomes on the Proton platform.

PGM in Production

Baylor's HGSC also has four Ion Torrent PGMs, a platform it first received in January of 2011 and moved into production mode about a year ago, using them primarily for amplicon and regional capture sequencing applications.

The center currently generates more than a gigabase per run on the PGM, using the 318 chip and 300 base pair reads. Applications include amplicon sequencing, mitochondrial genome sequencing, regional capture sequencing, exome sequencing, as well as 16S metagenomic sequencing and mate-pair library sequencing for structural variation detection.

Baylor has also started to automate library construction for the PGM, using standard Ion Torrent library protocols and Beckman Coulter Biomek FXp dual-arm robots, and has developed an Ion Torrent-specific single nucleotide variant caller called VarIONt.

One application of amplicon sequencing on the PGM has been to validate variants discovered on other sequencing platforms. Over the last two months alone, more than 4,000 variant sites from cancer projects were validated on the PGM, Muzny reported.

Another use of amplicon sequencing has been in disease discovery. For example, the center has screened in about 50 individuals, a large gene involved in a form of cardiomyopathy, which revealed 10 deleterious mutations in nine of the subjects.

In addition, the HGSC has tested a mitochondrial genome sequencing assay on the PGM that currently runs on the Illumina platform in Baylor's Whole Genome Laboratory. The assay uses long-range PCR and can detect mitochondrial heteroplasmy if variants are present in at least 10 percent of reads.

To validate this assay, Baylor compared data generated on the PGM to data generated on the Illumina MiSeq, using the same Charcot-Marie-Tooth syndrome test sample they used for their Proton comparison. They required 10 nanograms of input DNA for the long-range PCR reaction and 1 microgram of PCR products for library construction. Both the PGM and the MiSeq discovered the same 31 mitochondrial variants, Muzny said, but the PGM also found one true variant that the MiSeq missed, and the MiSeq data contained three false positive variants. In five additional previously analyzed patient samples, the PGM also discovered all known disease mutations.

More recently, the Baylor team has been developing regional capture designs as well as exome sequencing for the PGM. Optimization steps for the platform include reducing the amount of input DNA for library construction to one microgram and other changes to the library prep process.

So far, Baylor has designed a number of regional capture panels, ranging in size from about 0.1 to 7 megabases, for example for retinal disease, thrombosis, and parts of the X chromosome. It has also evaluated the VCRome 2.1 whole exome design on the PGM.

In terms of cost for running the PGM, Muzny said that the platform is currently "much cheaper" than the Roche 454 and "on par" with the Illumina MiSeq.