This is an archival copy of the Visualization Group's web page 1998 to 2017. For current information, please vist our group's new web page.

Visualization of Growth Curve Data from Phenotype Microarray Experiments


Contents

                          

Introduction

Phenotype is defined as the the observable physical or biochemical characteristics of an organism, as determined by both genetic makeup and nvironmental influences. Phenotype microarrays (PM) provide a high-throughput approach to screening the response of an organism to thousands of chemical treatments. The chemical treatments are delivered via substrates in microwells on the phenotype microarrays. Currently, phenotype microarrays contain either 24 or 96 microwells. A small sample of inoculum, containing cells of the organism being studied, is put in each microwell. An instrument called an Omnilog provides a digital readout of the change of opacity in each microwell. The change in turbidity (opacity) is used as a surrogate for cell growth.

The traditional way of looking at cell growth is to plot cell density (or a surrogate for cell density) vs. time (see adjacent figure). Under normal conditions in a microwell (i.e., limited source of nutrients), cell growth shows a sharp increase, reaches a maximum, and then levels off as cell increase is offset by cell death. When an organism is subjected to a stressor (e.g., high pH, high salinity, antibiotic, etc.), its growth is affected. The result is poor or no growth, a delay in the onset of growth, or a combination.     

Phenotype microarrays are available from Biolog, the maker of the Omnilog instrument. Biolog currently markets 20 phenotype microarrays that fall into different 'modes of action' categories. These categories include different sources (carbon, nitrogen, sulfur, phosphorus), nutritional supplements, nitrogen utilization, osmotic sensitivy, toxicity, pH, inhibitors, and chemical sensitivity. For example, on a phenotype microarray belonging to the carbon source mode of action category, the microwells contain different types of carbon compounds and different concentrations of the compounds. Given the range of modes of actions, phenotype microarrays may be used to look for the reaction of an organism to environmental stressors such as low/high pH, lack of nitrogen, high salinity, etc., or the effect of an antibiotic or growth inhibitor.

In a typical single-organism PM experiment, a researcher will inoculate several sets of 20 pnenotype microarrays. More than one set of phenotype microarrays is used in order to provide replicates as a means of judging the quality and reproducibility of the data. Within a few days, the PM experiment will have generated hundreds to thousands of cell accumulation curves (i.e., surrogate growth curves), depending on whether 24-well or 96-microwell phenotype microarrays were used.

Visualization Technique Used

In order to provide a more compact representation of the data from PM experiments, we display each cell accumulation curve as a thin horizontal line in which the magnitude of turbidity (opacity) is represented by color. Using this visualization technique, and by vertically stacking the horizontal lines representing the growth curves, it is possible to display cell accumulation data for all microwells on a single phenotype microarray in a relatively small space.

The adjacent figure shows cell accumulation data for a 96-microwell phenotype microarray. Each of the nine rows (labeled A through H in the figure), on the phenotype microarray contains twelve microwells (numbered 1 through 12). Time increases from left to right.

The adjacent color image corresponds to 96 of the plots above. In terms of space required, displaying the cell accumulation data for 96 microwells using the type of plot above requires nearly 200 times the space of the color image representation of the same data. Using the color image representation means that cell accumulation data for a dozen phenotype microarrays - the equivalent of over 1,000 curves - can be displayed on a single page.
    

Examples of Use

There are several ways in which the color map representations of phenotype microarray data benefit the researcher. Quality control is one important benefit. Because of the compact nature of the data display, a researcher can quickly scan the results from a set of 20 phenotype microarrays and be able to detect microarrays for which

                        
Example of abiotic reaction. Example of instrument malfunction.

In addition, by comparing color images for replicate runs (technical or biological), the researcher can gauge the overall quality of results. If the replicates compare well, then the researcher can feel confident that comparison of the phenotypic responses of two organisms is valid. The figure below shows four color images. The two on the left are technical replicates of a phenotype microarray for wild-type (occurs in nature) Desulfovibrio vulgaris Hildenborough (DvH), a soil-borne bacterium. The two color images on the right are technical replicates for a mutant of Desulfovibrio vulgaris Hildenborough. Because the images for each pair of replicates are similar, it is reasonable to conclude that overall, for many of the treatments on this particular phenotype microarray, the growth of the wild-type is reduced compared to that of the mutant strain.

Web Delivery of Data and Images

Though the emphasis in this page has been on the approach to displaying cell accumulation data, the color images are made available to researchers via the Web. A database is used to store information about PM experiments, and a Web interface (figure below) allows the users to select the experiments they are interested in (upper left screenshot). The small color images corresponding to the selected experiment(s) are then displayed (top center screenshot). If the user clicks on one of the color images, a larger version is displayed (top right screenshot). In the larger version, moving the cursor over individual color lines (which are slightly separated) pops up the plate and well identifier, the chemical treatment (well substrate), and the mode of action.

If the user clicks on one of the horizontal lines in the larger view, then a separate window opens that shows a single growth curve corresponding to that line/microwell (center bottom screenshot). If the user clicks on one of the row labels (left side of the large color image in the top right screenshot), then a page opens that shows growth curves for the twelve microwells in that row (bottom left screenshot). The user also has the option to download the data for a single microwell from the single-plot display page (center bottom and bottom right screenshots).
Currently, information for nearly 100 PM experiments has been entered into the database. Some experiments are single-plate experiments, and some include data from all 20 phenotype microarrays. Information about the PM experiments and the corresponding color images are available to researchers via the Web interface described above. Using the color image visualization technique, together with the database and Web interface, have enabled researchers to assess experiment quality and compare phenotypes of different organisms much faster than was previously possible. This software system has been in use for about three years.

References

J.S. Jacobsen, D.C. Joyner, S.E. Borglin, T.C. Hazen, A.P. Arkin, E.W. Bethel. Visualization of Growth Curve Data from Phenotype Microarray Experiments. In 11th International Conference on Information Visualization IV 2007, Zürich, Switzerland, July 4-6, 2007. IEEE Computer Society Press. LBNL-63251.

Sharon Borglin, Dominique Joyner, Janet Jacobsen, Aindrila Mukhopadhyay, Terry C. Hazen, Overcoming the Anaerobic Hurdle in Phenotypic Microarrays: Generation and Visualization of Growth Curve Data for Desulfovibrio vulgaris Hildenborough, Journal of Microbiological Methods. Accepted Manuscript. In Press.