Table I. Most relevant physiochemical peptide properties.
Rank | Correlation Coefficienta | Peptide property | Property typeb |
---|---|---|---|
1 | -0.53 | Peptide mass | Size |
2 | -0.36 | Average relative preference value at C1 (28) | Structural |
3 | -0.33 | Average activation Gibbs energy of unfolding, pH7.0 (29) | Hydrophobicity |
4 | -0.27 | Average hydrophobicity coefficient in RP-HPLC, C4 (30) | Hydrophobicity |
5 | -0.20 | Average normalized frequency of zeta R (31) | Structural |
6 | 0.20 | Average linker propensity from 1-linker data set (32) | Structural |
7 | 0.16 | Average hydrophobicity coefficient in RP-HPLC, C18 | Hydrophobicity |
8 | 0.15 | Average AA composition of EXT2 of single-spanning proteins (33) | Structural |
9 | -0.14 | Average normalized frequency of α-helix in all-α class (34) | Structural |
10 | 0.08 | Average relative population of conformational state A (35) | Structural |
11 | 0.07 | Average surface composition of AAs in intracellular proteins of thermophiles (36) | Structural |
a Peptide properties were iteratively selected from a pool of 550 total properties based on their Pearson's correlation with the intensity ranks in the training data set. Properties are sorted based on the absolute value of the correlation coefficient, which is an indication of their importance for classification. Negative correlations indicate inverse relationships. As each feature was selected, redundant features with interproperty correlation coefficients >0.3 were removed.
b Peptide properties were loosely categorized into three types, those corresponding with peptide size, secondary structure, and hydrophobicity.