November 1, 2013 (Vol. 33, No. 19)
Oliver Leven Ph.D. Head of Professional Services Genedata
New Capabilities Advance Target Discovery and Validation
High-content screening (HCS) is a powerful technology that addresses a variety of questions within biological research. Established for pharmaceutical small molecule screening, it is regularly applied in target discovery and validation. Capable of observing phenotypic changes, HCS requires a technically complex infrastructure of hardware and software components. This complexity is compounded by the management of HCS images and image-analysis results, which can be resource- and time-consumptive.
Genedata Screener® for HCS overcomes these challenges (Figure). Its integrated image-management capabilities handle all data-management issues even for heterogeneous, multi-instrument environments. And, its data-analysis functionality gives scientists access to a broad range of screening applications, completely independent from the high-content infrastructure.
The following outlines key HCS lab activities and challenges, and how Genedata Screener for HCS addresses them.
1. HCS Instruments and Image-Analysis Software
The breadth of HCS biological research is reflected by the variety of HCS instruments in laboratories. While each instrument comes with its own image-analysis application, additional commercial or open-source applications are also available; screening scientists must select the right combination of instrument and image-analysis software most appropriate for their research. While scientifically optimized, this workflow is not easily supported as integrating different software components is a manual process.
Genedata Screener for HCS tackles this workflow issue by automatically importing images from any instrument, preserving the original files. The results from image analysis are imported automatically as well and are annotated and linked to the images from which they were derived.
2. Cell-Level Data Analysis
HCS image analysis works on individual biological objects such as cells or nuclei: During segmentation, individual biological objects are identified; subsequent quantification provides a set of numbers describing them. Individual cell-based data must be aggregated to deliver a result per well as all cells in a well are treated the same and any difference is due to different phenotypes or biological variability.
For example, in a receptor internalization assay, the amount of receptor moving from the cell membrane to the nucleus may be of interest. During image analysis, information such as size and intensity, and nucleus size and intensity are obtained for each cell. To obtain a result the cellular information is averaged across all cells in a well. Whether a cell is included in the average is defined as part of the image-analysis process, before the actual data for the complete experiment is available. If you want to change selection criteria afterward, the image analysis must be repeated.
Genedata Screener for HCS stores cell-level data in parallel with well-level data enabling scientists to define and edit—on-the-fly—cell populations. It also allows scientists to generate new well results by selecting different cell aggregations from a wide range—from simple average-up to advanced methods such as Kolmogorov–Smirnov.
3. Plate Data Processing: Quality Control and Result Generation
In HCS experiments, the plate is the processing unit of interest. Each plate contains samples plus controls to gauge the signal range and for normalization. Scientists should perform plate QC to identify and invalidate suspicious wells based on one or more QC features; and to calculate quality metrics such as Z´ or Signal over Background. Calculation of new signals based on a linear composition of other measured features may also be needed. All necessary data processing and review is provided by Genedata Screener; basic operations are done in seconds, allowing scientists to focus on experiment QC.
Result calculation follows plate processing; for different experiments different results have to be calculated: In genetic screens, the validity of a connection between a known target and a phenotype is verified. In small compound screens, the potency of known substances is compared across different cell lines to understand the relevance of the biological system for the target–phenotype relationship. Compound combination experiments may be conducted to understand the mechanistic relation between the phenotypic effects of two compounds affecting the same target. Genedata Screener provides a single platform for these and all other types of screening experiments.
4. Image Access
Image access is vital in all HCS. A promising well requires visual inspection to ensure phenotypic changes are as expected and to exclude artifacts. Poor-quality wells are inspected to understand issues and to improve operational quality for subsequent experiments. In a small screen of about 1,200 compounds, there are 10,000s of images, making it impossible to review all of them. With Genedata Screener image management, scientists can display images for wells and compounds of interest with a single mouse click without leaving the software.
5. Results Storage, Access, and Browsing
With Genedata Screener, you can trigger the automatic transmission of all results for a selected experiment to a corporate data warehouse. Most data warehouses, however, cannot accept data from HCS (multiple results per well), cannot store HCS images (needed to validate the numerical results), and cannot handle multiple substances per well (as in compound combination experiments). Genedata Screener overcomes this issue with a result database.
This result database stores all relevant data in functional tables (e.g., a table for all compound potency results, one for all compound combination results, and one for all gene activities). These tables contain links to underlying data-analysis sessions in Genedata Screener, allowing scientists to interpret a given result in the context of the original data-analysis session. Furthermore, the result database stores the references to the wells of interest, for which relevant images can be retrieved by the stand-alone image viewer. Stored data can be browsed with a variety of downstream tools such as Spotfire, Excel, or web interfaces.
6. Image Life-Cycle Support
Image handling, a concern in all HCS experiments, is particularly important in target validation. Storing all images in full-resolution for long-term instantaneous access presents a serious investment in central hard-disk capacity. Genedata Screener image management supports archiving—for all image formats—while maintaining full access to the images in context of the downstream analysis or for result storage.
At image import time, one or multiple thumbnails are automatically generated for every image and separately stored from original image files. If the latter have been archived and sequentially deleted from the fast disks to free resources, thumbnails are still available and will allow scientists to operate the data-processing pipeline connected to image management without any noticeable difference. In the event that scientists want to re-access the original images, they can restore these at any time.
The aforementioned key HCS lab activities illustrate how Genedata Screener effectively eliminates multiple software packages and provides a systematic and standardized methodology for innovative HCS data management and analysis. Using Genedata Screener, lab scientists report the time it takes to conduct an experiment can be reduced from eight weeks to one week, with HCS data-handling times reduced by more than 85%. These time efficiencies and productivity gains are especially important for target discovery and validation groups, which run a diverse portfolio of complex HCS-based experiments. Genedata Screener for HCS streamlines data management and analysis for these groups, giving scientists instant access to images and empowering them to focus on the wet work.
Oliver Leven, Ph.D. (firstname.lastname@example.org), is head of Genedata Screener Professional Services/Genedata.