Applying NI LabVIEW Technology in Drug-Discovery Process

  Print Print

"The flexibility and feature-rich functionality of LabVIEW and the Vision Development Module, along with the seamless integration of our software with the vision and motion control hardware, helped us to successfully deliver the Crystal Finder System in a short time period."

- Michael D. Coleman, Ph.D., Coleman Technologies

The Challenge:
Searching for and identifying protein crystals present in “hanging drops,” which are suspended in the wells of microtiter plates, as part of the drug discovery process.

The Solution:
Creating an automated crystal finder system, including an imager and classifier, using National Instruments LabVIEW Express and Vision Development Module, along with NI image acquisition and motion control hardware and driver software.

Michael D. Coleman, Ph.D. - Coleman Technologies

The field of protein crystallization is an important component of the drug-discovery process. Proteins under investigation are mixed with various combinations of reagents in an attempt to discover a recipe that will create conditions suitable for the formation of protein crystals, which can then be examined via X-ray diffraction. The number of possible permutations of mixtures can reach the millions, making the search for the optimum recipe tedious.

At Coleman Technologies, Inc., an NI Silver Alliance Program Partner, we were contracted by a major pharmaceutical company to develop a system to automate the search for crystals in microtiter plates, with up to 96 drops of solution per plate. The software that we developed consisted of two separate modules, the crystal finder imager and the crystal finder classifier. The imager controlled the image acquisition and scanning of microtiter plates full of drops, while performing automatic adjustments for focus, positioning, lighting, and lens zoom control. The classifier analyzed the images and scored the drops based on crystalline and particulate content.

Imager and Classifier Operations and Functionality

The Crystal Finder Imager hardware consisted of a Pentium IV, 3.0 GHz PC equipped with an NI PCI-7334 stepper motion controller, a four-axis MID-7604 stepper motor driver, and a PCI-1428 image acquisition board. A Pulnix TM-4000 black–and-white digital camera supplied high-resolution (2000 by 2000) images. A Danaher, 12 in. travel, X-Y table positioned microtiter plates under the Thales-Optem 16X motorized zoom & focus lens system. The system also had two stepper-controlled filter wheels with polarizing material and RGB filters, which allowed for the acquisition of cross-polarized and color images. Light intensity from the LED light source was controlled via RS232.

The imager system could run in an automated remote mode, where commands were received over a TCP/IP connection, or in a manual mode. In manual mode the operator could interactively control all the system hardware, including focus, zoom, light intensity, polarizer wheels, and X-Y position. The operator could also configure numerous system setup parameters.

In remote mode, the imager responded to numerous commands and status queries from the microtiter plate-scheduling software, which also controlled a robotic arm that transferred one of up to 500 plates stored in a “hotel” onto one of four “nests” on the imager.

When commanded to scan a microtiter plate, the imager parsed an XML-based scan file, which provided information about what drops to image and the image acquisition parameters. The imager moved the X-Y table to the approximate location of each drop and then performed an iterative process to center on the drop, and adjusted the motorized lens magnification so that the drop filled the image and was in focus. The imager then acquired the necessary images for analysis according to the XML scan file.

Because the drop depth could significantly exceed the depth of field of the lens system at higher magnifications, a “slicing” routine was implemented. The routine took a number of images in different focal planes throughout the drop depth to guarantee that the crystal and particulate matter were in focus in at least one image.

All of the images from a microtiter plate were automatically saved and named based on the plate bar code, the current date and time, and other parameters such as the drop number. When a scan was complete, the plate-scheduling software transferred the folder of images to a dedicated PC running the crystal finder classifier. The classifier provided a set of three scores for each drop image – a drop finding score, object score, and crystal score. Each of these scores varied from zero to 1,000, with 1,000 representing the highest confidence.

The drop score indicated the confidence that the complete border of the drop was known. The location of this border was important for accurately determining the drop content. Once the classifier identified the drop border, proprietary image analysis developed with the LabVIEW Vision Development Module was used to identify any objects in the drop. When we found the objects, we used the system to determine if they were crystals or simply precipitate or debris. Determining the drop content was extremely challenging due to the fact that the drop could act as a lens, causing extreme intensity variations and dark rings around the drop border.

The classifier determined the object and crystal scores and returned all the image scores in a text summary file. Where a number of slices were taken off a drop, the classifier determined the slice with the highest crystal score.

Over a period of several days or weeks, the plate-scheduling software repeatedly commanded the imager to scan a microtiter plate for crystals. The classifier crystal scored, along with visual inspection, were used as the basis for determining which drops and crystals to harvest for further study with X-ray diffraction.

System Increases Accuracy While Saving Time and Money

The flexibility and feature-rich functionality of LabVIEW and the Vision Development Module, along with the seamless integration of our software with the vision and motion control hardware, helped us to successfully deliver the Crystal Finder System in a short time period. This system will accelerate the drug discovery process and save our clients money because they can automate the analysis of samples to occur 24 hours a day, seven days a week. In addition, computer analysis will lead to more accurate and consistent results, which will be helpful for trending and data analysis.

Author Information:
Michael D. Coleman, Ph.D.
Coleman Technologies
16 Regency Plaza
Glen Mills, PA 19342
United States
Tel: (610) 459-9646
Fax: (610) 459-9649

Bookmark and Share

Explore the NI Developer Community

Discover and collaborate on the latest example code and tutorials with a worldwide community of engineers and scientists.

‌Check‌ out‌ the‌ NI‌ Community

Who is National Instruments?

National Instruments provides a graphical system design platform for test, control, and embedded design applications that is transforming the way engineers and scientists design, prototype, and deploy systems.

‌Learn‌ more‌ about‌ NI