At last week's High Content Analysis Conference in San Francisco, vendors presented new software and collaborative efforts designed to help scientists more deeply mine data from high-throughput workflows, and to gain added value from experiments and better manage, analyze, and store data flooding from high-content screening experiments.
One alliance, with IDBS and PerkinElmer, will integrate image analysis with data management in the two companies’ platforms. Separately at the conference, PerkinElmer launched its new benchtop instrument, a high-content screening system called Operetta with built-in image-analysis tools and software “building blocks” to allow researchers to create workflows.
Also, Genedata announced its High Content Analyzer, a new module for the firm’s Genedata Analyst high content data-analysis platform, and presented HCSDiscover, a new data-mining product developed in collaboration with Thermo Fisher Scientific’s Cellomics.
The software integration announced by PerkinElmer and IDBS, a drug discovery and development software platform developer based in Guildford, UK, is aimed at giving researchers simultaneous access to data in PE’s Columbus image data-management system and IDBS’s ActivityBase XE screening platform.
Integrating Columbus and XE will “smooth the workflow” in research, ActivityBase XE product manager Andy Vines told BioInform last week. The collaboration is drawing to a close, he said, and IDBS is looking to release the next version of XE, version 7.2, sometime during the first half of this year.
XE allows scientists to analyze and visualize a variety of high-content analysis experiments such as enzyme and multiplex analyses, and this new version will have “the PerkinElmer integration capabilities,” he said.
Technically the integration is between the Open Microscopy standards database technology, OMERO, upon which Columbus is built, and the XE technology, which is .NET, said Vines. OMERO is an image-management and -analysis platform developed through the Open Microscopy Environment, formed in 2001, to promote universal file formats in imaging.
OMERO is a java-based server/client system that can be used to analyze micrographs and their associated metadata. [BioInform, Dec. 5, 2008].
Customers have been requesting data management structure to handle images, Martin Daffertshofer, global software product manager at PerkinElmer Cellular Technologies, Germany, told BioInform sister publication Cell-Based Assay News last week.
With the launch of its Columbus high-content screening data-management system in October, PerkinElmer also saw the need to integrate information from chemical libraries and high-content imaging. “We saw that the best solution currently available to do that was the ActivityBase solution from IDBS,” he said.
Being an open standards database and not a file server “differentiates [the PerkinElmer system] from other products that are out there at the moment,” Vines said. With a file server, he said, it’s “relatively easy” to pick up a link to a set of image files. Although “you have to know where to look in that database,” he said, it associates images and metadata more clearly. He said the Columbus system is a “much better way of managing high-content imaging.”
Capturing images with a file server system creates “a very large file server with lots of directories and lots of images,” Vines said. “The thing becomes just a little bit messy,” since the relationship between images and associated metadata are not well-defined.
With this impending integrated system, Daffertshofer said, high-content screeners are getting “a central repository for all kinds of images, regardless of origin,” he said, adding that that it supports PE’s platform as well as data from GE and Thermo Fischer instruments. Engineers from both companies worked on the project, he said, based on an agreement signed late last summer.
Although in theory integration would be possible without an agreement in place, “in this case it would have been very difficult to do without an agreement. Because Columbus is a very different type of image-management technology over and above what others have to offer at the moment,” Vines said.
Vines said his firm was looking to pursue this same kind of software collaboration with other vendors, but it was too early to offer specifics about who the next partners will be. “Suffice to say, they are the other players in the high content instrument business,” he said.
IDBS’ ActivityBase product has 200 customers, some of whom are performing high-content screening experiments that include image generation in large pharmaceutical companies, biotech, and contract research organizations with its primary customer base in pharma and biotech, Vines said.
Scientists derive results and numeric values from a given image, analyze those numeric results, and then make decisions around the experiment based upon those numbers and the analyses of those numbers, Vines explained.
“Generally speaking, the image files tend to be completely non-integrated with the results from the analyses of data that are derived from them,” he said. “Normally the images go somewhere and the data go somewhere else.”
The idea behind the collaboration between the two companies, he said, was to give scientists “that integrated view of both the images that are generated and the data and results that are generated from those images.”
Getting the systems to mesh has mainly involved integration, not new programming. “It is relatively straightforward for us to use our application interface to connect to our partner application interface and therefore provide an integrated solution to an end-user,” he said.
IDBS has experience pulling data from a variety of instruments into ActivityBase, he said, but images add an “extra layer” that would be lost if the companies had decided to just use the current method in which just the image analysis results flow into ActivityBase rather than both the analysis and the images.
“What we are doing is providing that link to the original images from within ActivityBase. If you are doing analysis for the first time, you can immediately access the images and the data that you have generated. If you go back to it, you can continue to access the images along with the results,” Vines said.
A Seat on a Benchtop
At the HCS conference PerkinElmer also launched a benchtop instrument called Operetta for high-content screening and analysis that is designed to “allow maximum functionality in the smallest possible footprint,” the company said in a statement announcing the release. The tool contains a “scalable” database for image management as well as data analysis and retrieval.
As Daffertshofer explained to BioInform sister publication Cell-Based Assay News, Operetta is an “entry-level” instrument, “the little sister to the Opera,” the company’s confocal micro-plate imaging reader used in automated high speed and high resolution screening and which applies image analysis software called Acapella. Operetta’s optics and image-resolution capabilities are “pretty close” to Opera’s, Daffertshofer said.
Operetta’s typical users are likely to be scientists already running HCS experiments who may be involved in cellular assays but not necessarily imaging, he said. These users “see that there is an added value to imaging and are looking for a system that makes it convenient to work it. “This, I think, is the most important market for the Operetta system,” he said.
Compared with Opera, Operetta has “modified functionality in terms of the software,” said Daffertshofer. It is powered by PerkinElmer’s new Harmony software, which he calls “the bridge between the various components” of the instrument. PerkinElmer said the software offers users a set of computational building blocks to create workflows, letting them develop high content assays, and giving users remote access as well as image-analysis and -storage options.
Users can develop their own algorithms using these pre-existing building blocks, he said, with one component working the hardware, another the image analysis, and the third directing the database.
“Coming up with a very powerful data-analysis tool just takes you a few clicks with the software,” Daffertshofer said.
The idea with Operetta, he said, was to simplify software use and make image capture easier. “Simplicity was of utmost importance when building the Operetta system,” he said.
The instrument also contains “ready-made solutions for frequently used image-analysis tasks.” It has preconfigured image-analysis applications and such features as image segmentation and building blocks it creates customized image analysis.
Jumping Around in the Data
Meantime at HCS, Genedata made two debuts: the new High Content Data Analyzer module and HCS Discover software co-developed with Thermo’s Cellomics.
Researchers scaling up their experiments, and thus their workflow, and who are current users of the firm’s screening data-management and -analysis platform Screener are, “in relatively good shape” as they, for example, seek to adapt the system they use for high-throughput screening for high-content screening, Stephan Heyse, general manager for screening informatics at the Switzerland-based company told BioInform in an e-mail last week. The standard Screener system “is able to handle the additional layers of data and increased data set size quite easily,” he said.
Other researchers face a situation in which their system doesn’t scale, he said.
“It was not designed to capture more than one read-out per well and they are faced with a series of workarounds to handle the multiple layers in HCS data,” he said. Their current system is also “strained” when required to measure 1 million or fewer wells “with only one measurement per well.”
“These systems simply don’t function when they try to run these same systems with 10 or 20 measurements captured per well,” said Heyse.
Describing the firm’s rationale for propelling development of new software such as High Content Analyzer and HCSDiscover, Heyse explained that he sees that most HTS customers are also doing HCS. “We have watched this market grow with interest and have developed a series of smaller improvements in functionality up to this point,” he said. “Given the requests for specific functionality in the growing HCS arena, we have come up with a module release dedicated to HCS functionality.”
Genedata wants to help customers better mine their HCS data and one aspect — “a very simple part,” he said — is to enable them to interactively view HCS images while performing analyses and making decisions as they generate hit lists in drug discovery.
This type of concurrent image viewing is currently possible with Screener, he said, and the new High Content Analyzer module improves this capability. As an add-on module to the existing platform, called Assay Analyzer, the High Content Analyzer offers visualization tools, access to HCS images, and tools “to more easily make use of multiple HCS read-outs.”
“Part of this improvement is to bring the viewing of images more within the general framework of Screener, while another part is the creation of a new application programming interface, so that Screener can work directly with the whatever HCS vendor they are working with,” Heyse said.
“The first example of this” is the collaboration with Thermo Fisher and the integration with its Cellomics Store, a relational database designed to help scientists organize HCS data and image analysis, he said.
In the past, users needed to export files between systems, Heyse said, explaining that the new High Content Analyzer eliminates the need to have an output file as an intermediate step. “This direct connection eases this data exchange,” he said. Output files have been standard with HTS readers, but “the data volume and complexity is greater with HCS.”
The new module has built-in statistics and visualization tools so users can navigate through their data, assess their data quality, and make decisions based on that analysis, he said.
Genedata Analyst, another tool presented at the HCS conference, is biostatistics software for large-scale mining of data from HCS experiments, and includes adding in the context of gene expression, proteomics, and metabolomics data, which can be analyzed along with clinical outcomes.
High Content Analyzer is part of the company’s Screener software platform that integrates data streams from different instruments and screening technologies along the drug discovery pipeline, Heyse said.
To give Screener the possibility of interfacing with that range of HCS systems from different vendors, “Genedata has spent a great deal of effort developing a sophisticated and well-documented application programming interface for bringing in these data,” he said, adding it involves Jython, a java-compliant scripting language written in Python.
Genedata offers to help customers with scripting, but he said that most users manage this “one-off customization” on their own. “Once a data parser or integration is created, it is available to all end users via a graphical user interface,” he said.
At the conference, Genedata also announced a new data-mining product developed in collaboration with Thermo Fisher called HCSDiscover slated to be a Cellomics high content screening and informatics product. The product will be launched in March of this year, Mark Collins, Thermo’s marketing manager for high-content cellular imaging and analysis told Cell-Based Assay News.
HCSDiscover, said Heyse, “uses the Genedata Screener infrastructure but is packaged and tightly coupled with the Cellomics platform, enriching Cellomics' HCS informatics portfolio with a powerful module for downstream data analysis and mining.”
The software is targeted at low to medium throughput experiments and at individual HCS laboratories/researchers, while Genedata’s High Content Analyzer is “an enterprise solution for HCS, from low- to ultra-high-throughput [labs],” Heyse said.
A Round Environment
Heyse said that Genedata sets itself apart from competitors in the HCS space with its decade of experience “in methodology and software for handling and analyzing complex and large-scale biological data.” He also sees an advantage in his firm’s “solutions and data integration options beyond HCS in the complete biological space,” including pathway analysis and toxicology experiments based on HCS data.
Vines compared the IDBS/PerkinElmer partnership to the Genedata offering. IDBS’ products handle analysis and data management, making results “accessible to the wider corporation [and] the wider customer base. … That differentiates us because Genedata doesn’t have the underlying data management and storage capabilities,” he said.
The data is captured in a structured Oracle database, he said, and “therefore become available to other parts of an organization and make[s] it easy to integrate.”
IDBS has clients active in GxP environments, he said, the compliance arena of good manufacturing practice, good laboratory practice, and good clinical practice, all of which are scrutinized by Food and Drug Administration. The IDBS product offering reaches from early discovery into GxP environments, he said.
“We are seeing a number of customers taking our products into GxP environments, so we’re providing the additional pieces of work, validation in the software, that gives them that,” Vines said.