Table 2. Time Performance Benchmark Results of the Different Grammars Applied to Different Lipid Name Test Setsa.
| test set | avg len | C++* | Java* | Python* |
|---|---|---|---|---|
| Goslin (S) | 15.59 | 8237.23 | 7074.57 | 2957.81 |
| Goslin (L) | 37.76 | 2880.18 | 7771.71 | 807.31 |
| LIPID MAPS | 38.57 | 3727.17 | 5498.62 | 996.36 |
| SwissLipids | 36.84 | 3691.4 | 5810.29 | 973.80 |
| HMDB | 32.41 | 4266.21 | 9100.97 | 1280.73 |
a
All values for the programming languages C++, Java, and Python 3 have the unit parsed lipid names per second*. All test sets contain 10 000 lipid names parsable by their grammar. Here, the test set Goslin (S) contains rather short lipid names on average and Goslin (L) rather long lipid names. The parsing time is dependent on the length of the lipid string. Therefore, the average length (avg len) over all lipids in each data set is reported. Note, that the R implementation utilizes the C++ library internally. Therefore, it is not listed here explicitly.