bootf2BCA_v1.4
GUI version
Copyright (C) 2016-2025 Aleksander Mendyk
This program is distributed “AS IS” in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
Contents
1. Introduction......................................................................................................................2
2. Requirements...................................................................................................................3
3. How to run.......................................................................................................................3
4. Input/output data.............................................................................................................. 8
5. Citation.............................................................................................................................9
6. Changelog........................................................................................................................ 9
7. License........................................................................................................................... 10
page 1 of 24
1. Introduction
This program was developed to assist with f2 computation in cases when intra- and inter-
batch variability is large, namely RSD>10%. The use of statistical bootstrap technique
allows to implement confidence interval (CI) of the f2 coefficient resulting in overcoming
of their major drawback in the original metrics. The algorithm provides possible “worst
case scenario” of f2 values, thus supporting claim about equivalence of in vitro
dissolution profiles. Several types of CI are calculated, among them bias corrected and
accelerated (Bcα) and recently advocated by EMA percentage-based. Three statistics are
available for calculations: (1) f2, (2) unbiased f2 estimator and (3) expected value of f2.
Target users are researchers from industry and academia dealing with comparison of
dissolution profiles. The software is Open Source and developed with R statistical
environment. This is an extension of my former project PhEq_bootstrap also available
from the sourceforge.net website.
2. Requirements
R environment (https://www.r-project.org/) with additional packages installed:
• boot (https://cran.r-project.org/package=boot),
• shiny (https://cran.r-project.org/package=shiny),
• shinyFiles (https://cran.r-project.org/package=shinyFiles)
• plotly (https://cran.r-project.org/package=plotly)
You may use install_prerequisites.R script to install the above packages in your
environment.
3. How to run
For command line interface go to the CLI directory. Put there both reference and test
products files with dissolution data. Next edit the section “user-set parameters” of
bootf2BCA_v1.4.R script. After that simply run the script bootf2BCA_v1.4.R
interactively or non-interactively with your R environment. GUI is reflecting all
controlling parameters of the original script thus there's no need to edit script manually.
For GUI you may either use shiny’s command runApp() from within root directory,
where the R scripts are located or alternatively you can use Rstudio to run this application
as well (www.rstudio.com). In the first mode the app will run with in the browser, yet
locally thus no communication with any external server is required. Therefore, no
information is being sent out whatsoever.
Right after the start the GUI looks like presented on Figure 1. There is a navigation bar at
the top of the window and several controls on the left panel. A dummy data based graph
of the dissolution profiles is displayed. Next step is to provide reference and test data
with dissolution profiles. Please push the buttons “Browse...” in the left panel of GUI:
“Reference file” and “Test file” respectively. Once the files are loaded and validated for
their consistency, a new graph is created, this time with the loaded data. The validation of
the reference vs. test data is brief and checks only whether the timepoints are the same.
After this stage the GUI should look like on the Figure 2. Windows users will be
restricted to the root directory of the software itself to access data from and save data to.
This is dictated by Windows security policy and is also relevant to the GUI version.
There is an option to work with many subdirectories.
Next stage is to tune the algorithm parameters using controls from the left panel:
• Report file (F2_boot.R script variable: output_file_report)
This is the report file name to be generated for logging all results. The default
value is "report.txt".
• No. of bootstraps (F2_boot.R script variable: bootstrap_number)
Number of bootstrap runs. The default value is 1000.
• CI range (%) (F2_boot.R script variable: CI_probability_range)
Confidence interval probability in percentage. The default value is 90
If necessary you may tune additional parameters from the “Advanced options” of the top
navigation bar
• Q>=85% auto cut-off rule (F2_boot.R script variable:
my_85_percent_auto_rule)
This variable controls automatic procedure for cutting of the profiles when
crossing the threshold of 85% dissolution. Available options are: 0 - never use it
(never adjust the profiles); 1 - whenever one of the profiles exceeds 85% of
dissolution then cut both the profiles, regardless of which one is the first one; 2 -
only when both profiles exceed 85% then cut their terminal parts
The default value is 2
• Sampling mode (F2_boot.R script variable: sampling_mode)
Two sampling modes are introduced for bootstrapping: 1 - individual values
drawn from the pool of each time-point, meaning each time point is treated
page 5 of 24
separately and the profiles are generated "de novo"; 2- whole profiles as blocks
for bootstrapping menaning that bootstrap samples are created from the original
profiles only - no "de novo" generated profiles are introduced.
The default value is 2
• Boot simulation type (F2_boot.R script variable: boot_package_simulation_type)
Options of boot package are available for more sophisticated bootstrap scenarios:
"ordinary" and "balanced". For more detailed understanding please refer to the
boot package manual (https://cran.r-project.org/web/packages/boot/boot.pdf)
The default value is "ordinary"
• Statistic (F2_boot.R script variable: calculated_statistic). Selection of statistic to
be bootstrapped. One of the three available: f2, unbiased f2 and expected value of
f2. Default is f2 (calculated_statistic = 1)
• Confidence intervals (F2_boot.R variable boot_ci_type). By default all types of
CI are enabled (my_boot_ci_types = c("norm","basic","perc", "bca")). CI “bca”
should not be used with unbiased f2 statistic.
• Seed (F2_boot.R script variable: my_seed) setting value of seed for pseudo-
random numbers generator
• Number of decimals (F2_boot.R script variable: round_digits) – number of
decimal places to round the numbers to. The default value is 2
After setting up the parameters please push the button “Run calculations” in order to run
the algorithm. Please bear in mind that depending on the “no of bootstraps” parameter the
calculations can take from seconds to hours on a modern computer. Usually there’s no
need to exceed 10 000 bootstraps.
4. Input/output data
Two files are required to perform calculations: reference and test file representing as per
FDA terminology the dissolution results of reference and test formulation respectively.
These files are tab-delimited with each dissolution profile represented in a separate
column, where the first column is a time axis. No labels are provided in this file, neither
for columns nor for rows. The dissolution results are expressed as percentages of the
label claim
An example of such file is presented below and additionally ref.txt and test.txt files are
available in the root directory for testing purposes
30 36.1 33 35.7 32.1 36.1 34.1 32.4 39.6 34.5 38 32.2 35.2
60 58.6 59.5 62.3 62.3 53.6 63.2 61.3 61.8 58 59.2 56.2 58
90 80 80.8 83 81.3 72.6 83 80 80.4 76.9 79.3 77.2 76.7
180 93.3 95.7 97.1 92.8 88.8 97.4 96.8 98.6 93.3 94 96.3 96.8
The script logs all relevant data into the report file that is a txt file suitable for importing
to the spreadsheet like i.e. MS Excel® or LibreOffice Calc.
5. Citation
If you publish with this software please cite following paper:
Aleksander Mendyk, Adam Pacławski, Jakub Szlęk, Renata Jachowicz. PhEq_bootstrap:
an Open Source software for simulation of f2 distribution in cases of a large variability in
the dissolution profiles. Dissolut Technol, 20(1), 2013, 13 - 17.
6. Changelog
v1.4.
• added calculation of unbiased estimator of f2
• added choice dialog of calculated confidence intervals
Download Latest Version
bootf2BCA_v1.4.zip (1.2 MB)
