January 1, 2008 (Vol. 28, No. 1)
Researchers Attempt to Make Sense of Massive Amounts of Data
High-content analysis (HCA) is emerging as one of the fastest growing sectors in drug discovery and development. It represents the convergence between cell-based assays, high-resolution imaging, and advanced image processing and analysis. Making sense out of the massive amounts of data can be daunting. New tools are aiming to improve multiple parameter analysis to enhance data management and to provide more intuitive visualization methods.
Scientists using imaging systems generate thousands of images from which they measure millions of parameters. There are three major limitations for customers performing high-content screening (HCS), according to Pierre Turpin, Ph.D., product application scientist, Molecular Devices (www.moldevices.com). Dr. Turpin is one of a number of scientists who will speak at Cambridge Healthtech’s “High Content Analysis” conference in San Francisco later this month.
“The first challenge is how to organize data. This can be a significant problem because scientists need to compare many experiments that each generates a massive amount of data. The second challenge is how to analyze all of the data. The measured parameters need to be analyzed at once with multiparametric methods, which are difficult to understand and use. The third challenge is how to connect the results of analysis and the original image taken by the imaging system.”
Molecular Devices developed its AcuityXpress cellular informatics software as a complete solution for high-content data analysis and visualization, reports Dr. Turpin. “Most of the software products available are limited. AcuityXpress solves each of the three issues by using a seamless integration to allow better organization, providing lots of tools for multiple parameter analysis, and allowing visualization of the original image.” One of the features of AcuityXpress is the MDCEarth™ module, which melds visualization and high-content data.
“It’s a unique and groundbreaking tool that allows direct interaction of images and numerical data at all levels and in both directions. It’s like Google earth. You can see a demo on our website (www.acuityxpress.com). Following a large screen, zooming in provides a detailed view of individual images on a plate down to the level of the individual cell. You can also zoom out to a more global, high-level view of all image data from a large screen and see overall patterns. It’s quite fascinating to look at your data in this way. You can derive information you could otherwise miss entirely.”
Molecular Devices also offers a complete solution for HCS with a new generation of ImageXpress imaging systems, which use image analysis based on the MetaMorph platform and the MDCStore database.
Maximizing the value of screening data is critical in the environment of fast-paced data analysis and decision making. “Screening experiments produce a lot of data, and the volume is increasing all the time. It is important to have quality data because at the end of the day, you need to find leads,” says Glyn Williams, vp marketing, IDBS (www.idbs.com).
IDBS offers its ActivityBase suite of tools for drug discovery data management. “ActivityBase is data-management software that captures data from instruments and immediately transfers it into a database. For high throughput, you really want seamless integration that provides real-time analysis,” says Williams.
“Additionally, the XE module is designed to streamline workflow for ultra-high-performance analysis. This provides an extensive amount of tools specifically for data quality and consistency. Thus, you can use the same environment to capture, visualize, and verify data. It also has a flexible and intuitive ability to visualize your data. This is usually not available in most spreadsheet packages.”
Another feature researchers need, according to Williams, is flexibility to perform multiple analyses on plates. “We incorporated a free-form plate format that allows scientists to perform a variety of assays on the same plate such as single-point primary screens and dose-response secondary screens. Additionally, customers can customize methods and build their own templates.”
HCA and Genomic siRNA
HCA can underpin target selection, and target selection represents the first critical step in the process of small molecule drug discovery, according to Laszlo Kiss, Ph.D., research fellow, automated biotechnology, Merck & Co. (www.merck.com).
“Mining the primary sequence of the human genome to identify novel targets that can be positioned for full-scale drug development is a challenging, multidisciplinary process. Integrating the most predictive in vitro methodologies for correlating the presence of a molecular target with a disease-relevant phenotype is the gateway to novel target identification and validation. Our aim is to improve the process of how we select targets for development and thereby reduce costly attrition later in drug development.”
Techniques used to control gene expression including both loss of function (RNAi) and gain of function (cDNA) are readily accessible for high-throughput screening (HTS) at a genome-wide scale. These techniques are used principally for functional genomics, the analysis of protein function in cells. “We have applied these technologies for functional screens in order to gain a deeper understanding of disease associated pathways so that we may validate, or invalidate, molecular targets for drug discovery,” notes Dr. Kiss.
Merck uses a number of technologies such as confocal, widefield, brightfield, and light-scatter microscopy for HTS. “High-content imaging offers a number of advantages for functional genomic screening,” explains Dr. Kiss. “Imaging permits the temporal and subcellular spatial tracking of proteins in living cells. In addition, it provides multiplexed readouts such that we can integrate responses of multiple cellular targets and processes at the primary screen stage. Imagine also allows us to identify and analyze cell subpopulations making physiologically relevant cell types (e.g., stem cells and primary cells) accessible for screening.”
Dr. Kiss’ team has utilized RNAi and cDNA to manipulate protein expression levels and combined it with high-content imaging to measure phenotypic function at the cellular level in an effort to probe complex cellular pathways. “These studies are conducted at genome-wide, druggable-genome, and focused-genome scale,” he says. “We are beginning to see examples where we have identified novel genes that function in well-established biological processes.”
Viewing the cell as the basic unit of life and therefore the first useful level of systems biology, Cellumen (www.cellumen.com) focuses on a cellular systems biology approach for drug discovery, development, and clinical trials.
“The day of the blockbuster drug is over or certainly limited,” says D. Lansing Taylor, Ph.D., CEO. “We now need to begin addressing the complexity and diversity of humans, and cells can be the surrogates of that complexity. Cell-based assays are the fastest growing sector of drug discovery and development.”
Cellumen is initially targeting hepatocytes in safety profiling since more than 50% of drug toxicities are liver based, says Dr. Taylor. The company assess both human and rat primary cells.
“Rat cells are included since the FDA mandates animal testing and primary rat cells can be used to filter compounds in the development pipeline. It’s important to do this early on in the process so we have a filter for selecting lead series and with fewer failures at later stages. Ultimately, the goal is to have a predictive tool for human toxicity.”
Cellumen scientists employed an 11-parameter, multiplex cytotoxicity panel in a recent study involving 130 compounds with safety data. “We measured three different time points in our model: one for an acute response, one hour; one for an early measure, 24 hours; and one as a chronic measurement, for example 72 hours, for compound treatments,” explains Dr. Taylor.
“The 11 parameters included mitochondrial function, oxidative stress, and cell cycle. The entire concept is that we look at a systemic readout of functional biomarkers. This generates a massive amount of data so we then couple this to informatics. The end result is that we can rank compounds and create a safety index for predicting hepatotoxicity during the investigational safety phase.”
In addition to their own programs, Cellumen offers custom profiling services, informatics, access to its databases, and safety assessments for clients who wish to outsource. The first panel of reagents and protocols for human hepatocyte profiling is available from Millipore in a comarketing arrangement.
Most scientists are not computer programmers and vice versa. Bridging that gap is a necessary challenge for HCA. This has been a priority for GE Healthcare Life Sciences (www.gelifesciences.com) in designing its software for high-content analysis, data mining, and visualization.
“We offer a high level of intuitiveness in our software products,” says Ger Brophy, Ph.D., GM of GE Healthcare Life Sciences Advanced Systems. “Our intent is to keep all users as up to date as possible for high-content analysis. In the past, this was difficult due to the complexity of the software and the fact that scientists or programmers needed to write their own code or develop their own analysis algorithms. Scientists need more deductive tools. The real challenge currently is to provide software solutions to manage and analyze data in a faster and more convenient environment.”
The company is collaborating with EMC to provide extensive database management and mining through the In Cell Miner HCM product, which will be released in the first quarter of 2008.
“Our software is developed for biologists by biologists,” reports Dr. Brophy. “Miner also interfaces seamlessly with our data visualization and analysis software, for example, our Investigator offering. One of the most popular features of the Investigator product is multitarget analysis. This allows the user to measure a wide variety of cellular parameters and classify cells using interactive filters to identify relevant clusters and subpopulations. This ability is especially important when working with heterogeneous populations such as those present in stem cell research.”
These products will provide comprehensive tools for understanding cellular relationships in high-content analysis, indicates Dr. Brophy. “We also couple these tools with a high level of support. For example, we can offer remote support for assay development and problem solving through our Bio InSite™ digital services delivery platform.”
HCA is making strides in targeting difficult cells such as neurons. Researchers at Elan Pharmaceutical (www.elan.com) are using Cellomics’ (www.cellomics.com) BioApplications image analysis tool to measure synaptic function in primary neurons. BioApplications provides researchers with measurements that discern biological properties of the cell such as size, shape, amount of label, and pattern of fluorescence. BioApplication reports include cellular data on individual cells as well as whole-well and subpopulation levels.
Elan scientists investigated a panel of small molecule neurotoxins and observed that different classes of neurotoxins demonstrated differential effects on parameters of neuronal morphology. For example, they determined that phosphatase inhibitors had a greater effect on branch points than on neurite length.
On the other hand, using 6-OH dopamine affected all parameters of neuronal morphology equally. The company reports that its study demonstrates how the use of high-content analysis assays can allow monitoring of neuronal morphology and function. This could ultimately lead to discovery of new drug candidates for therapies.
HCA is poised to continue as a key player in drug discovery. The level of capital that companies are expending shows confidence in their expectations that new technologies will continue to expand the area’s potential.