Common cause of pipeline crashing late into a run is that there is an error in the path to databases in the config file. One way to improve this would be to check these files exist when the pipeline is launched (not within a rule) Could also check that all of the input reads files exist