Visualization and Curve-Parameter Estimation Strategies for Detailed Exploration of Phenotype MicroArray Kinetics
The DSMZ authors were motivated to extract as much information as possible from the high throughput phenotyping provided by Biolog Phenotype MicroArrays. They hypothesized that multiple kinetic parameter estimation is necessary to capture the diversity of the biological response recorded over time. Therefore they adapted R functions and existing statistical methods for modeling kinetic curves into a package that is customized to work within the biological context of Biolog PM kinetic redox dye reduction curves.
Data was created using Biolog's phenotypic test plates and the OmniLog® PM System. Using the R functions levelplot and xyplot, they implemented methods to display kinetic curves as thin lines to improve the display of multiple curves. Using R's add-on package grofit, they compared model fitting and free spline fitting approaches to parameter estimation. The DSMZ authors used technical replicate experimental design to evaluate the utility of parameter estimation and confidence interval display. They also used archetypal curve analysis and showed that two to four archetypes were necessary to represent the range of positive test reactions. These new methods of dimensional and noise reduction improve the ability to detect statistically significant differences in kinetic curves that are within the defined limits of biological reality. The authors suggest that these methods and tools may be used for automated post processing of Biolog PM Data.
Link to PDF file of this publication
Link to the fact sheet and first release of OPM R Package with tools for analyzing OmniLog® Phenotype MicroArray (PM) data including:plotting
aggregating (estimating curve parameters)
comparing and discretizing PM data
creating phylogenetic formats and reports for taxonomic journals
integrating metadata, using the YAML format for the storage of data and metadata
batch conversion of large numbers of files
customized data frames for application of other R packages
A user-friendly manual is available from Dr. Markus Göker (email@example.com) or Dr. Johannes Sikorski (firstname.lastname@example.org) upon request.