Q&A and Troubleshooting

Nicolas T. Wirth

2023-04-04

This Troubleshooting guide aims to provide assistance and solutions to common issues that users may encounter while working with the QurvE package or shiny app. As we gather more user feedback, we will continue to update this page with additional information and solutions for a smooth QurvE experience.

Dealing with Outliers

Outliers can significantly affect the accuracy and reliability of growth curve analysis. Identifying and addressing outliers is an essential step when using QurvE for high-throughput growth curve experiments. In this section, we discuss some strategies to identify and handle outliers in your data.

Identifying Outliers

1. Visual Inspection: Use the “Validation” window in the Shiny app to visually inspect each growth or fluorescence curve and its corresponding fit. Look for any irregularities or deviations from the expected growth pattern.

2. Examine Growth Parameters: Plot the computed growth or fluorescence parameters (e.g., growth rate, lag time, and maximum OD, fluorescence increase rate) for each sample to identify any unusual values that may indicate an outlier sample.

3. Bootstrapping: QurvE offers bootstrapping for spline fitting, which can help estimate uncertainty, validate the model, and identify potential outliers. Use the bootstrapping results to assess the reliability of your curve fits.

Handling Outliers

1. Modify Fitting Parameters: Adjust the fitting parameters in the [Computation] section of the Shiny app to systematically exclude problematic sections of the measurements (e.g., outside the detection limit of the instrument). In R, this can be achieved by defining values for t0, tmax, min.growth, and max.growth in the growth.workflow() and fl.workflow() functions

2. Re-run outlier samples: After identifying outliers, re-run their analysis with modified parameters in the [Validation] panel of the shiny app. An example for how this can be achieved within R is provided in the growth curve analysis documentation

3. Data Preprocessing: Before importing your data into QurvE, preprocess it to remove or reduce the impact of outliers. This can be done using various data cleaning techniques, such as filtering, interpolation, or imputation. Please note that such preprocessing methods are outside the scope of QurvE.

Q: What is the recommended sampling interval for different fitting methods?

A: The appropriate measurement interval depends on the organism and growth conditions being studied. For instance, fast-growing organisms like Escherichia coli require shorter sampling intervals of at least one hour while slow-growing organisms like plants can be measured up to every five days. It’s also important to note that the appropriate sampling interval may vary depending on the specific research question being addressed. For example, if the goal is to study the effects of a particular treatment on growth rates or diauxic shifts, more frequent sampling may be necessary to capture the short-term changes in growth. Alternatively, if the goal is to assess overall growth patterns over a longer time frame, less frequent sampling may be sufficient.
For linear fits, we recommend having at least four measurements within the mid-log phase of the growth curves to reliably estimate growth rates. When using smoothing splines with a low measurement frequency, such as in bioreactor fermentations or shaken flask experiments, you can increase the smoothing factor to make the fit less susceptible to outliers.