Skip to main content
Premium Trial:

Request an Annual Quote

Entering Life Sciences Market, Appistry Reworks Cloud Offering for NGS Data Analysis

Premium

By Uduak Grace Thomas

Cloud computing firm Appistry this week threw its hat in the bioinformatics ring with a version of its CloudIQ platform that provides a complete data-analysis pipeline for next-generation sequencing data.

Dubbed Ayrris/Bio, the new product marks the first offering from a dedicated life-sciences group that the company put in place earlier this year to develop and commercialize technologies for genomic data analysis.

The group adds to teams that the company already had in place for the defense and financial services markets, which offer the industry-specific Ayrris/Defense and Ayrris/Finance offerings, respectively.

Appistry CEO Kevin Haar told BioInform that the company's initial introduction to life sciences began a year ago with requests for its solutions to handle genomic data. At the time, he said, Appistry didn't know what to expect from the field. At best, he said, the company intended to provide a complete cloud solution for scientists, researchers, and analysts. At the worst, it hoped to come up with some general improvements for its platform.

However, over the last six to nine months, the company realized that the genomics marketplace was a "fantastic fit" for its technology and required more than just passing attention, he said. Around the same time, Appistry began assembling its life-sciences team, culminating in the launch of its cloud solution for the space.

The St. Louis-based company currently has several dozen employees. Haar declined to provide a specific headcount for the company or the life-science group but said the firm is currently looking for computational biologists and computer engineers as well as sales and marketing staff.

In addition to looking to market its CloudIQ platform for specific vertical industries, Appistry renamed it in an attempt to give it a "fresh positioning" and clear up some perceived confusion associated with the overuse of the term "cloud," Haar said.

In addition, he noted that the new branding is intended to indicate that the platform is not just a simple cloud architecture, but is rather an analytics framework capable of meeting clients' unique needs.

"Cloud computing has gotten to be a little bit of everything to everyone," he said. "It's gotten pretty 'cloudy' as to what cloud computing really is, so there's been less attraction to that kind of positioning for us."

For life sciences customers, Ayrris/Bio's framework makes it easier to "plug solutions into our environment without programming," Haar said. He added that the system is amenable to a lot of "relatively standard analytic pipelines ... [used] to analyze genomic data, whether it's exome or whole-genome pipelines or RNA-seq."

Haar expects that Appistry's life-science arm will soon outpace the company's legacy defense and finance businesses, estimating that it could constitute 50 percent of the firm's business by the end of this year.

Inside Ayrris/Bio

Appistry's life-sciences group is responsible for supporting Ayrris/Bio as well as developing new technologies and approaches meant to enable researchers to build, test, and deploy sequencing pipelines.

Among other capabilities, Ayrris/Bio supports several sequence and analysis programs and provides support for analytics algorithms such as Monte Carlo simulations, MapReduce, and graphical model inference.

It also includes analysis pipelines for human exome data; whole-genome data for human, bovine, maize, and other organisms; and RNA-seq data.

Appistry plans to add additional pipelines to the platform but Haar declined to provide specifics.

Customers can either purchase the platform as an appliance that can be installed in house or pay to use the firm's public cloud service, in which Appistry's team analyzes the data and returns the results to the client.

Its exome analysis service is priced at $750 per exome while its whole-genome service goes for $1,000 per genome. The company also offers volume discounts to companies who run multiple samples per month.

The company offers three types of appliances with a medium appliance providing about 1.2 terabytes of storage, a small appliance offers about half that capacity and a large appliance that provides over 2 terabytes of storage. The lab version of its platform, which is its smallest appliance, starts at under $100,000.

Appistry expects its pricing scheme will compete favorably with offerings from more established cloud providers in the space.

These include Eagle Genomics; DNAnexus; Geschickten, which also recently renamed its NGS analysis cloud platform (BI 6/3/2011); Geospiza, which was recently acquired by PerkinElmer (BI 5/6/2011); and other newcomers to the space like Samsung SDS, which began beta testing its cloud-based NGS analysis service this month (BI 8/26/2011).

It's unclear whether Appistry will win out on price since other companies use different pricing models. DNAnexus, for example, offers academic customers a rate of $20 per gigabase of raw sequence, which includes analysis and storage for two years. At that price, a whole human genome at 10-fold coverage would cost $600, but that would increase to $1,800 for 30-fold coverage.

"Our strategy is that we are really going to drive the price/performance of this market space," Haar said. "Because of our architecture we can do things faster and less expensively ... and at a higher quality so we are not asking people to sacrifice anything."

He noted that in trial runs against traditional grid technologies such as SunGrid Engine and Platform Computing, Appistry's infrastructure offers 10 times the performance on one-tenth the infrastructure because of its "computational storage" capability, which combines storage and execution systems.

Unlike traditional clusters, Ayrris so-called "computational storage" feature uses information about file locations to guide incoming work requests to the machine holding the relevant data files. Files are read at disk speed instead of network speed so that latency is reduced and data throughput is increased, according to the company.

"What that means is that we can bring the processing request to where the data is already local so we don’t spend any time waiting for [input/output] ... all those other grid technologies or other homegrown strategies are trying to push this data around they are bringing it off of storage facilities into a computational environment and then sending it back," he explained. "We don’t have to do that, we process it right in place and the net result is we are much faster."

The company claims that a medium-sized Ayrris appliance can process a human exome — comprising 184 million reads of 100 base pairs — in 34 minutes and a whole human genome — comprising 567 million reads of 100 base pairs — in 97 minutes.


Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.