CombiStats version 5.0 was released on 11 March 2013. Upgrading from version 4.0 to version 5.0 is optional and comes at no extra cost. The software is backward compatible so all your existing datasheets and templates will work as before. Your current password will also work for version 5.0. This is a quick guide to the new features:
- Password protection of templates and datasheets
Many users asked for the possibility to protect templates and datasheets from accidental editing. This is now provided via a new button with a padlock icon . There are 4 levels of protection and each level can have its own password. Passwords are optional. If you type no password when increasing the level of protection, the user can remove the protection without password. Even if no password is used this provides an increased safeguard against accidental modification. More information on the 4 levels of protection and the use of passwords is provided in the manual.
- Support for equivalence testing
Equivalence testing can be activated in the options wizard’s tab ANOVA. CombiStats will then show a table with equivalence statistics. Shown are the slopes of the individual preparations and their 90% confidence interval (not 95%!). If a standard is included the difference and ratios of the slopes with that of the standard are also shown, again with 90% confidence limits. For slope ratio models the intercepts are used instead of the slopes and in that case no ratios are calculated since this has no meaningful interpretation. The purpose of these statistics is to check whether the complete 90% confidence interval is contained within pre-defined “goal-posts”. If this is the case, it is concluded that the slopes are equivalent, even when there is significant non-parallelism according to the traditional ANOVA. Here is an example.
- Robust regression to reduce the effect of outliers
Although the European Pharmacopoeia discourages users to apply outlier tests to exclude outliers, it has been noted that many laboratories still apply such a test blindly without looking into an experimental explanation for the occurrence of these values. To encourage these laboratories to abandon this practice in favour of better alternatives version 5.0 offers robust regression by the use of Huber’s weights. This can be activated in the advanced options by specifying a weight of w=h. Here is an example.
- Automatic invocation of Spearman/Kärber method
Some assays that are designed to be analysed with probit models (or similar parametric curves), may have insufficient data in the central part to allow for estimation of the slope. In these cases an estimate can often still be obtained using the Spearman/Kärber method. In version 4.0 it was then necessary to change the model specifications from probit to angular with fixed slope of 1/ln(dilution-step). This is no longer necessary in version 5.0. CombiStats detects when the parametric models cannot be used and automatically invokes the Spearman/Kärber method. A message that this is the case is printed on the output. Here is an example.
- Faster convergence of non-linear models
A faster algorithm is implemented in version 5.0 so many cases where slow convergence occurred in version 4.0 will now converge much quicker. Unweighted non-linear models in particular will converge faster than before, especially where asymptotes are difficult to estimate. The new algorithm should produce the same outcome as the old algorithm but in rare cases the last decimal may differ by at most 1.
- 5-parameter sigmoid curve regression
This was requested by some users. It allows for a 5th parameter to model asymmetry of the curves. It can be activated in the options wizard’s tab “transformation” and is available only for quantal models and quantitative sigmoid models. Here is an example.
- Improved UNDO/REDO functionality
A new button to REDO an action that was reversed by UNDO is available. Another improvement is that the UNDO functionality in version 4.0 could cause flickering of the screen which was the cause of sometimes slow data-restoring. This inconvenience has been solved.
- Graphs connecting mean responses
The full-page graph has a new button to generate a graph that connects the mean responses per dose. Here is an example.
- Faster computation of the Wilcoxon-Mann-Whitney statistic
The algorithm for the computation of the exact WMW statistic was further optimized so that larger sets of data can be analyzed within a reasonable lapse of time. Very large problems (say more than 20 experimental units per sample) may still be too slow for exact computations in which case the BREAK button (or CTRL+ESC) provides an approximation.
- Ratios in single dose assays
In CombiStats 4.0 it was not possible to enter ratios of the form r/n for single dose assays based on quantal responses. The only workaround possible was to enter a series of r zeros and n-r ones. This is no longer necessary in version 5.0.
- Quick access to example files
Some users had difficulties locating the folder with example files, or were completely unaware that examples are installed with the software. A new button with a light bulb icon now provides direct access to the more than 60 example files.
- Toggle display of mean responses on and off
When mean responses are displayed by using CTRL+R (or the corresponding menu item), it was previously necessary to edit something in the data sheet to make it disappear again. In version 5.0 you can hide the mean responses by simply repeating the action CTRL+R.
- Exporting matrices
It is now possible to export matrices when calculations were interrupted with the BREAK (or CTRL+ESC) key. This was not possible in previous versions.