Skip to main content

Table 1 Attributes used in machine learning algorithms.

From: Improved machine learning method for analysis of gas phase chemistry of peptides

Set A: 27 Attributes derived from sequence
AacidN1, AacidN2, AacidN3, AacidN4, AacidN5 First 5 amino acids on N-terminus.
AacidC1, AacidC2 Last 2 amino acids on C-terminus.
HydnN1, HydnN2,..., HydnC2 Hydrophobicity for each of the above seven amino acids.
BasicityN1, BasicityN2,..., BasicityC2 Basicity for each of the above seven amino acids.
Ave_basicity Average peptide basicity.
NumRs Number of arginine residues in peptide.
mobileH Number of basic residues subtracted from peptide charge (indicates existence of mobile proton).
NumHKR_RN2 Number of basic residues to the left of N2 bond.
NumHKR_LN2 Number of basic residues to the right of N2 bond.
OMW Observed Molecular Weight.
Set B: 5 Attributes derived from MAE feature recognition function
OYMinusB The balance between y and b ions.
NumIon Total number of ions.
P_intensity Intensity of the parent ion.
Osum (sum of intensity of observed major ions)/(sum of intensity of all ions in the MS/MS output).
Tsum (sum of intensity of theoretical major ions)/(sum of intensity of all ions)
Set C: 6 Attributes based on scores generated by database search/sequence validation programs and from our sequence validation methods
Mowse Mascot's score
Xcorr Sequest's score
SumScore Summary score; A combination of Sequest's XCorr and Mascot's Mowse score.
PIC Proportion of the total ion current score for each MS/MS spectrum which accounts for fragment ion assignments.
SIM Evaluates chemical plausibility based on relative fragment ion intensities when comparing observed MS/MS spectrum to theoretical spectrum.
InterScore The percentage of observed fragments accounted for by multiple fragmentation events.