Modeltest

Modeltest is an advanced and reliable piece of software developed to serve genetics engineers in determining the differences between DNA substitution models using hierarchical hypothesis testing in order to generate the results.

The application is fairly complex, as it is meant for experienced users with a solid background knowledge in genetics as well as computers, specifically since it does not provide users with a GUI, running only through the command line interface, which can make it fairly challenging for novices.

Nonetheless, Modeltest does not require installation, so it suffices for users to just drop the executable file onto an open instance of Command Prompt and press the Enter key, which in turn will reveal the program's functions along with a brief explanation of the available commands.

Modeltest is meant to work with files generated by PAUP* which contain the genetics information required, as well as files which comprise raw log scores along with parameter estimates for the assessed models.

On the opposite end, the utility generates a description of the probability ratio tests that were conducted, and their corresponding P-values, information which is then interpreted and assigned a proper model.

The supported models are listed in the command line interface, including JC (Jukes and Cantor), TrNef (Tamura-Nei with equal base frequencies), TIM (Transitional Model), GTR (General Time Reversible), F81 (Felsenstein) and several more.

At the same time, users can work with a specific set of functions, exemplified in the main window in order to make them easier to understand and employ. As such, users can determine the debug level ( -d), the alpha level (-a), the number of taxa (-t), activate the LRT calculator mode (-l), determine confidence interval for averaging (-w) or use BIC rather than AIC for all future operations (-b).

To conclude, Modeltest is an efficient application developed for professionals working in the field of genetics, offering them the means of comparing DNA substitution models, thus helping users decide on which best fits their data.