HTTPS SSH

Data and analytical code for Allaway et al.

DOI

Written by: Casey S. Greene, Robert J. Allaway, and Balint Z. Kacsoh

In Vitro Immunofluorescence Intensity Analyses

See the manuscript for details on preparation, labeling, etc. After labeling measurement was performed in Fiji. The region was selected in the DAPI channel. Each measured region was paired with an adjacent background region. The source code begins with these data (measurements and background are on alternating) lines. Source code in combine_lines.py generates a summarized data table that includes background subtracted intensity values.

Plotting and analysis are performed via plot_cells.R.

Diagnostics

Our goal was to examine the total intensity of fluorescence per unit area. Because we're going to divide these two items, we want to make sure that a decrease in cell size after treatment, or some other similar effect is not driving observed changes across treatments or replicates. The figures shown here merge replicates, but identical diagnostic figures faceted by replicates are included as _facet in the repository.

We examined area across conditions.

area-vs-treatment

If we think that this is due to fluorescence intensity we should see differences in intensity across treatments, and we do.

intensity-vs-treatment

We also plotted bivariate relationships with each treatment to identify potential issues and everything looks fine here too.

bivariate

Result

We performed a number of analyses to examine potential changes in the form of the distribution (e.g. two fluorescence populations with a shift) and did not observe this (diagnostic figures in the figures/cells folder), so we showed a boxplot with notches as the summary figure:

boxplot

FNA-PDX Growth Curves During Dinaciclib Treatment

See the manuscript for details on the generation and characterization of the FNA-PDX models. Models were treated as described in the paper with Dinaciclib or vehicle control. Data are provided in raw-data/008-volumes.txt and raw-data/018-volumes.txt. Data are aggregated by process_volumes.py, and plotted by graph_volumes.R.

Results

We performed LOESS regression and calculated 95% confidence intervals. Growth curves are calculated for both FNA models as well as a patient matched models of metastases. The vehicle control is shown in red, while the Dinaciclib treated models are shown in blue.

FNA-PDX:

boxplot

Metastatic Models:

boxplot

We also generated figures using the traditional (mean +/- standard error) approach, and these are shown below.

FNA-PDX Mean +/- SE:

boxplot

Metastatic Models Mean +/- SE:

boxplot