Because we typically connect molecules in a dataset via a redundant network we can handle in principle failure to complete some perturbations (due to runtime errors, or due to large statistical uncertainties in free energy estimator).
The analysis step of the workflow should be fault-tolerant, and discard pathways that involve edges with unreliable/unavailable free energy estimates.
Because we typically connect molecules in a dataset via a redundant network we can handle in principle failure to complete some perturbations (due to runtime errors, or due to large statistical uncertainties in free energy estimator).
The analysis step of the workflow should be fault-tolerant, and discard pathways that involve edges with unreliable/unavailable free energy estimates.