Table 1.
The weights of the features in L-score reported by logistic regression using the ST training envelope matches
| Feature | #peak pairs =2 | #peak pairs =3 | #peak pairs =4 | #peak pairs ≥5 | ||||
|---|---|---|---|---|---|---|---|---|
| Weight | P-value | Weight | P-value | Weight | P-value | Weight | P-value | |
| M/z distance | 3.237 | 3.5E-4 | 4.987 | 2.0E-16 | 3.942 | 2.0E-16 | 3.820 | 2.0E-16 |
| Intensity distribution | 0.851 | 0.028 | 1.565 | 2.8E-8 | 1.448 | 6.2E-8 | 1.349 | 2.0E-10 |
| #supporting envelopes | -1.517 | 1.6E-5 | -1.471 | 8.7E-9 | -1.958 | 2.0E-16 | -0.992 | 2.0E-16 |
| #neutral loss envelopes | -2.491 | 5.5E-13 | -0.810 | 2.38E-7 | -0.795 | 2.8E-13 | -0.343 | 1.1E-11 |
| #missing peaks | - | - | 0.277 | 0.198 | 0.386 | 9.1E-3 | 0.096 | 0.314 |
When the number of peak pairs is 2, the weight for the number of missing peaks (feature m(E ′ )) is not used because this group does not contain any experimental envelopes with missing peaks.