Fig. 5
The effect of the number of hourly data points on the quality of parameter predictions as reflected by the R2 metric. Two case studies are shown: estimation of generalized mass action (GMA) parameters using the SRES algorithm (left column) and the Michaelis–Menten (MM) parameter estimations utilizing the G3PCX algorithm (right column). The top row and bottom row are for the cases without and with measurement noise, respectively. For GMA parameter predictions, the amount of measurement noise is sampled from a normal distribution with mean zero and a standard deviation equivalent to 5% of the underlying metabolite/enzyme measurement. The corresponding value is 7.5% for MM parameter predictions. The highest noise level has been chosen in each case such that the algorithm still performs well with 4 hourly datapoints. Triplicate values are derived using different initial seed solutions (Additional file 8: Data S7). The bars represent the median values while the lower and upper bounds (in red) mark the smallest and largest values based on the three initial seed solutions