Some errors are recoverable. For those errors, a Python ModellerError exception is raised. It is then up to your script to deal sensibly with the failure of the preceding command using a standard 'except' clause. For example, this flexibility allows derivation of multiple models and searching for many sequences, even if some cases abort due to convergence problems.
There are five kinds of messages that MODELLER writes to the log file: long output from the MODELLER commands, short notes to do with the execution of the program (files opened, etc.), warnings identified by `_W>', errors identified by `_E>', and the messages about the status of dynamic memory allocation. To control how much of this output is displayed, use the log object; for more information, see Section 4.14.