Fig. 9

Runtimes of the identification algorithm for the training and validation sets. The total runtime per compound is shown in black. The runtime of specific steps is also depicted: for the knapsack (Step 1), for the graph construction with all knapsack fragments (Step 2), for the enumeration of all minor-isotope chemical formulae above the LOD (Step 3), for the optimisation of contribution of sets of fragments, using a machine learning algorithm (Step 7). For most compounds, Step 7 remains the most time intensive step. The corresponding numerical values are given in Tables 6 and 7