San Diego, Calif.-based Edico Genome is preparing to launch its first product for the bioinformatics market, an application-specific integrated circuit system for next-generation sequence data analysis called the Dragen Wavefront Processor, which consists of a chip loaded with proprietary algorithms for mapping, alignment, sorting, and variant calling.
Edico plans to officially launch Dragen during this year's Bio-IT World conference to be held in Boston from April 29 to March 1. It is currently testing the system in conjunction with a number of undisclosed sequencing centers.
The company is positioning its appliance as an alternative to large server systems as well cloud infrastructure, which, in the case of the former can be costly to purchase, install, and maintain; and in the case of the latter may not mesh with the privacy requirements of some customers.
"Our goal is to get rid of those racks of servers and computers that are associated with the sequencing machine today," Pieter Van Rooyen, the company's founder and CEO, told BioInform this week, and to replace them with a single cheap chip — which is plugged into sequencers' control computers — that provides fast, efficient analyses for a fraction of current costs, without compromising accuracy.
According to the results of internal tests that Edico shared with BioInform, a standard Dragen configuration — which consists of a single chip mounted on a PCIe expansion card — takes about 20 minutes to analyze a whole genome from mapping through to variant calling. It's also able to compress and decompress large datasets rapidly, requiring about five or six minutes for the process, Van Rooyen said.
Other internal tests that compared the performance and speed of Dragen to a large cluster showed that while a single chip was able to analyze data from 150 genomes — generated on a set of Illumina HiSeq X Ten sequencers — in three days, the large cluster needed to use 52 servers to complete the same task in the same amount of time.
The company said that it has compared the proprietary algorithms — which are also optimized for chip architecture — that Dragen uses for mapping, alignment, and variant calling to standard software such as the Burrows-Wheeler Aligner (BWA). Its tests show, Van Rooyen said, that the Dragen algorithms are as accurate as, and in some cases do better than, existing tools. The company plans to publish a paper that will provide details of a comparison between its mapping and alignment tool and BWA. The results will show that it compares "very favorably" to BWA, Van Rooyen said, and that it does slightly better than BWA when aligning regions with SNPs and indels.
Edico isn't disclosing pricing for its product publicly, so it isn't clear how its price point stacks up against the costs of existing hardware options. Van Rooyen said that the company has adopted separate business models for its potential research and clinical clients. He declined to discuss the option for clinical customers, but he noted that research labs will be able to lease Dragen for use in their projects. He also said that these clients will be charged a fee per gigabase of data analyzed but declined to provide additional details about the lease terms and the data fee.
Edico also is keeping the exact details of the access options that it will offer to customers close to its chest, but clinical and research users will be provided with different versions of the Dragen technology. "Clinical use requires more accurate assessment of variants, and our clinical grade solution provides for an increase in accuracy," Van Rooyen said.
In an effort to drum up interest in Dragen chips, Edico is offering potential customers a chance to test drive a web-based version of Dragen on small quantities of sequence data prior to committing to a leasing agreement. This free version of Dragen will provide access to a subset of the capabilities that are available on the full system to give customers a sense of the speed and accuracy that it can deliver, Van Rooyen said. Users will be able to upload FASTQ files; run mapping, alignment, and sorting tools; then download the resulting BAM file for further analysis. After they've had a chance to test it, interested clients can then follow up with the company with regard to the costs and terms of implementing the system in their labs.
Dragen is currently optimized to work with data from Illumina's sequencing instruments but it can and will be extended to work with systems from other vendors in the future, Van Rooyen said.
Edico Genome officially opened its doors last year and is based in La Jolla, Calif. The 11-person company intends to double its headcount this year.