^{1}

^{2}

^{3}

^{*}

The authors have declared that no competing interests exist.

Conceived and designed the experiments: DN. Performed the experiments: DN. Analyzed the data: DN. Contributed reagents/materials/analysis tools: DN. Wrote the paper: DN.

In the process of new cancer drug development, as the first step of their assessment, their activities are usually studied _{50}): the concentration of the tested agent that inhibits the proliferation of the cancer cell population to 50% of the theoretically possible effect (absolute IC_{50}) or maximum effect practically achieved by the drug (relative IC_{50}). The currently available software for calculating IC_{50} values requires manual data entry, is time consuming, and is prone to calculation errors. Thus, we have developed open source, free, easy-to-use software for performing standardized data evaluations and automatically calculating the IC_{50}. This software eliminates the laborious and error-prone manual entry of data, substantially reduces the amount of time spent for data analysis. It has been extensively used in our department as the main tool for

High-throughput

The results of _{50}): the concentration of the tested agent that inhibits the proliferation of the cancer cell population to 50% of the theoretically possible effect (absolute IC_{50}) or maximum effect practically achieved by the drug (relative IC_{50}). The detailed description between the two types of IC_{50} values is provided in the materials and methods section. The data obtained during screening tests and the IC_{50} must be calculated manually, usually using spreadsheet programs. As an alternative, the same calculations can be performed using commercially available statistical software (GraphPad Prism from GraphPad Software) or the drfit package for the R Statistical Environment

We aimed to fill this gap and developed free, open-source, easy-to-use software that was specifically designed to extract data, calculate IC_{50} values, and produce a comprehensive report of the analysis. Our software, Cheburator, is user friendly, customizable, does not require programming or extensive statistical skills, and comes with detailed documentation. It can also process many text-based data files using its built-in convertor. This software has been extensively tested by our research group for several years and has substantially reduced the amount of time spent by our scientific staff on data processing. Because it is free for academic use, it can be successfully adopted by other research groups working in the area of

Cheburator was initially developed to process the data from sulforhodamine B (SRB,

This software was designed to analyze the results of assays performed on 96-well plates using the layout shown in

The sample plate with 1 control row (H), 1 row for the reference drug (A), and 4 rows for 4 novel studied compounds (B, C, D, and E, respectively) tested in 4 concentrations. Please note that not all rows have to be used on the plate, and different ranges of concentrations for different compounds can be applied in the test. Concentrations are highlighted in red. Control of untreated cells (CC) and control of media (CM) are in blue and green, respectively.

Depending on the option selected in the program (_{50} values _{50}, let us consider a theoretical dose-response curve where concentration of the studied compound is shown on the abscissa axis and percent of inhibition is shown on ordinate axis (_{50} is defined as a concentration of studied compound which results in exactly 50% of the maximum inhibition effect achievable in assay (dashed pink line in _{max}). Relative IC_{50} is defined as the concentration of studied compound which results in exactly half of the maximum inhibition effect attainable for that particular compound (dashed orange line).

(_{50} which can be estimated by the software. Absolute IC_{50} is defined as the concentration of tested compound which results in 50% inhibition of cell growth as defined by assay controls (pink lines). Relative IC_{50} is defined as the concentration of tested compound which results in half of the maximum inhibition of cell growth attainable for that particular compound (orange lines). I_{max} is defined as upper plateau of the dose-response curve (solid orange line). (_{50}. In the second method, a linear model is built with the intercept and the slope, which are calculated by linear regression analysis using data points which are >0% and <100% (red lines). These parameters are later used to estimate IC_{50}. In the third method, the software uses all data points to build nonlinear regression model and estimate IC_{50} (blue lines). Absolute IC_{50} values are shown on this plot, but software can also estimate relative IC_{50} in linear and nonlinear regression analysis.

The percentage of inhibition of cell proliferation in each respective well is calculated relative to untreated cells using the formula:_{cm} = mean background absorbance of wells without cells (control of the medium); A_{cc} = mean absorbance in wells containing untreated cells (control of the cells); and A_{p = }absorbance value of the well containing cells treated with the tested compound at the concentration of interest.

The percentages of cell proliferation inhibition are used to calculate the IC_{50} value. Cheburator supports three approaches – the simple two-points method previously described by

This is the simplest among the three methods implemented in our software and can only be used to calculate absolute IC_{50}. It assumes linear dose-response relationship in concentrations resulting in near 50% of inhibition and thus, its use is only recommended for preliminary assessment. First, the average percent of inhibition is calculated for each tested concentration (triplicate). Then, the IC_{50} value is estimated by linear interpolation using the 2 data points representing averaged percents of inhibition bracketing 50%. The IC_{50} value or its logarithm, depending on the dose axis transformation, can be graphically represented as the abscissa value for a point of the line drawn between 2 data points where it reaches the value of 50% on the ordinate axis. For example, suppose we have 2 concentrations, 10 and 100 µg/ml, that caused 25% and 65% inhibitions of proliferation, respectively _{50} value is calculated using the formula:

C_{lower} = concentration of tested preparation, which resulted in <50% inhibition of proliferation; C_{higher} = concentration of tested preparation, which resulted in >50% inhibition of proliferation; IP_{lower} = inhibition of proliferation (%) calculated for C_{lower}; IP_{higher} = inhibition of proliferation (%) calculated for C_{higher}.

If one or both data points bracketing the 50% inhibition of proliferation themselves result in 0% or 100% inhibition of proliferation, the IC_{50} value will not be calculated because of the uncertainty of the estimation. It is therefore reasonable to repeat the test for this compound using the narrower dose range, which would provide better data for properly estimating the IC_{50} value. Another case in which the IC_{50} value cannot be properly estimated using the two-points method is when little difference exists between the 2 data points bracketing the 50% inhibition of proliferation. The IC_{50} value would not be estimable if 1 or both data points bracketing the 50% inhibition were in the interval close to 50% (e.g., 48% and 52% by default). Due to the simplicity of the IC_{50} calculation in this method, program does not report 95% confidence intervals for its value in this case.

In this method of calculating the IC_{50} value, the linear regression line is built on the basis of all the data points for which the calculated percent of inhibition is more than 0% and less than 100% (solid red line in _{50} value (red arrow and IC_{50} value in _{50} calculation, it usually has to be combined with the transformation of 1 or both axes to ensure the proper conversion of the dose-response curve into the linear approximation. The abscissa axis (doses) can be analyzed in either logarithmic scale with the base 2, 5, and 10 (default) or linear scale. The ordinate axis (percent of inhibition), which is linear by default, can also be analyzed using probit, logit, or logarithmic transformations. Goodness of fit in linear regression analysis and therefore the reliability of IC_{50} calculation for each particular compound can be assessed using ^{2}_{50} value calculation, our program also reports 95% confidence intervals for its value, estimated using the semi-parametric bootstrapping method

This method is currently considered to be “state-of-the-art” for IC_{50} calculation _{50} calculation as it does not require linear relationship between drug dose and percent of inhibition and uses all data points from the test, unless they were specifically discarded from the analysis (see below). The program uses the _{50} value (blue curve, arrow and IC_{50} value in _{max} (a maximum attainable inhibition effect, solid orange line in ^{2} value, which is used to assess goodness of fit. The closer R^{2} value is to 1, the better the model fits the data and the higher the accuracy of IC_{50} estimation. In the case of nonlinear regression analysis, the program also calculates and reports 95% confidence intervals for IC_{50} value, estimated using the semi-parametric bootstrapping method

We recommend that the tests for each compound are repeated several times in independent assays, followed by the calculation of geometric mean of IC_{50} values and its 95% confidence interval. The geometric mean is statistically the most appropriate averaging method for IC_{50} values, due to the fact that they follow log-normal distribution

Cheburator was developed using the open-source

After the software has loaded, users can open a data file containing the results from a single

The line containing the data from the first row of the plate is selected in the picture. Other rows have to follow immediately after the first row. All rows before and after the plate data will be discarded. Each row is then divided into separate values using the delimiter specified in the combo box below (semicolon is the default, but other delimiters are also supported). Since each row could contain more values than the number of wells, the user also has to specify in the edit box below which value in the row will be treated as the optical density value for the first well. Data for the subsequent wells have to follow immediately after, separated by delimiters. Note that the user can use a period or comma as the decimal separator for data files. Cheburator will automatically convert all decimal separators into periods, which are the default.

When the data are loaded, the

The absorbance values in wells are highlighted with the intensities proportional to the ratio between their respective values and control values.

Next, scales for the _{50} (two-point, linear or nonlinear regression, with the latter set by default).

After specifying the relevant details for the test, users select the other plate rows that contain test data for the analyzed compounds using the check boxes located on the left side of the plate data table. The name of the compound and the dose ranges must be specified for every selected row using the

It is also possible to discard or modify single values from the plate. To do this, the user must double-click on the desired value in the plate table of the main window (

This window allows the user to change or discard particular values from the analysis. Any changes will be noted in the final report.

Changed values appear in red, and discarded values are crossed out after data modification.

After the analysis, 2 additional _{50} values (if applicable), their types (absolute or relative), 95% confidence intervals, goodness of fit values (linear and nonlinear regression), I_{max} (nonlinear regression), and the method of estimation are shown below the respective table. Triplicates for which percentages of inhibition of the cell population were calculated to be 0% or 100% are highlighted in red.

It is showing the dose range applied, mean and standard deviations of absorbance values for every triplicate, calculated respective percentages of inhibition of cell proliferation, estimated IC_{50} values, its type, method of calculation, and 95% confidence intervals (for regression analysis).

It is showing the dose-response curves, individual values of percentages of inhibition for each data point, method of calculation used, IC_{50} values, its type, 95% confidence intervals, and goodness of fit (for regression analysis).

User may examine the dose-response curves for each compound analyzed on the _{50} values. Here, individual data points are plotted in black, which helps to see the experimental variability between absorbance values from different wells. In cases when either the linear or nonlinear regression method is used for estimation, the regression line is also shown in the respective graph. Calculated IC_{50} values (if reached) are shown as green points on the graphs, together with a fuchsia line marking respective percent of inhibition for either absolute or relative IC_{50}.

Following the analysis, the user can print a detailed report that contains all relevant data: the path to the original file with the raw data, the file’s date and time, the name and date of the test, the calculated IC_{50} values, the percentages of growth inhibition for every concentration and compound, the dose-response curves for visual assessment, the mean optical densities and their standard deviations, the test control values, regression model lines and parameters (if applicable), and the type of the method used for calculation (

This sheet contains the original file’s name, date and time, the name and date of the test, the calculated IC_{50} values, its type, 95% confidence intervals, and goodness of fit (for regression analysis), the percentages of growth inhibition for every concentration and compound, the dose-response curves, the mean optical densities and their standard deviations, the test control values, and the type of the method used for calculation. A table with raw plate data is printed at the bottom part of the report.

Finally, there is also an _{50} calculation method can be specified (52% and 48% by default, as explained in the Materials and Methods section). _{50} type in regression analysis_{50} is calculated during the linear or nonlinear regression analysis. Users can also change the number of digits displayed after the decimal separator in the counted IC_{50} values. In addition, color highlighting of absorbance values can be printed in the final hard copy report if the

This tab allows user to customize different analysis and report parameters of the software.

Cheburator is open source and is available for academic use, but a license is required for commercial use. The latest version of Cheburator, together with its source code, sample data files, and documentation, can be downloaded from the program’s web page at

Manually estimating IC_{50} values is laborious, time consuming, and prone to calculation errors. Our Cheburator software program has been extensively used in our department as the main tool for

Other software exists that can perform IC_{50} calculations, for example, the commercial GraphPad Prism program (GraphPad Software) or the drfit package for R

If users prefer to fit the final dose-response curves with data in other software packages, our software is still useful for performing reliable preliminary assessments since it requires only minimal data entry and manipulation effort and has features that are useful for visualizing the results. For example, Cheburator could be used during the early screening stages to calculate the intermediate IC_{50} values, assess the approximate magnitudes of drug potencies, or quickly adjust the dose ranges for subsequent tests. We believe that this tool will be useful, either alone or combined with other software packages, for research groups working in the area of anticancer drug discovery.