Currently the input conditions are embedded in the process tag as a /-separated list, but this approach has proven to be very inflexible. Perhaps it would be better to specify conditions as key-value pairs in the process tag so that they can be arbitrarily ordered. The aggregate.py script can then extract them into separate columns.
The example pipelines for KINC and HemeLB each have several explicit processes that perform different types of experiments, but the script body for each process is essentially the same. We could probably reduce them to one process and provide an input file which simply lists the conditions for each experiment per line, also as key-value pairs. That way we could also do multi-dimensional parameter sweeps.
bentsherman
commented
4 years ago
Currently the input conditions are embedded in the process tag as a
/-separated list, but this approach has proven to be very inflexible. Perhaps it would be better to specify conditions as key-value pairs in the process tag so that they can be arbitrarily ordered. Theaggregate.pyscript can then extract them into separate columns.The example pipelines for KINC and HemeLB each have several explicit processes that perform different types of experiments, but the script body for each process is essentially the same. We could probably reduce them to one process and provide an input file which simply lists the conditions for each experiment per line, also as key-value pairs. That way we could also do multi-dimensional parameter sweeps.