Quantitative structure-property relationship studies for collision cross sections of 579 singly protonated peptides based on a novel descriptor as molecular graph fingerprint (MoGF)

Aiming at ion mobility spectrometry (IMS), computer-assisted ion mobility prediction (CAIMP) has been recently developed to simulate and predict diverse IMS behaviors in assistance of mathematics and computer science. Of that, quantitative structure-property relationship (QSPR) plays a vital role, dedicating to predict properties of unknown samples by creating statistical model based on known samples. In QSPR, the key lies in how to transform structural characteristics of target compounds into a group of numerical codes. In consideration that future IMS applications may mainly focus on intricate drug/biological systems, a novel molecular structural characterization method referring to molecular graphic fingerprint (MoGF) is proposed in this paper. In MoGF approach, radical distribution function is employed to map intrinsic interatomic correlations into a coordinate system according to a reasonable sampling interval, thus forming the characteristic graph curve which is rich in information on molecular structural characteristics, possessing of great merits in easy calculation, independent of experiments, large information contents, explicit structural meanings and intuitive expressions, etc. Consequently, MoGF is utilized to QSPR studies on 579 singly protonated peptide collision cross sections, and the constructed partial least square (PLS) regression model is confirmed to be robust and predictable by rigorous both internal and external validations, with statistics as r2 = 0.991, q2 = 0.990, RMSEE = 5.526, RMSCV = 5.572, qext2=0.990, rext2=0.990, r0,ext2=0.990, r0,ext'2=0.990, k = 1.003, k′ = 0.996 and RMSEP = 5.561, respectively.