Computerized Assignment of Near-IR Absorbances to Molecular Motions of Proteins and Peptides
Elizabeth G. Kraemer
Factor analysis of near-IR spectra of proteins and peptides in samples often reveals 20 or more possible proteins above the spectral noise level, even in in vivo near-IR spectrometry. One approach to understanding changes in near-IR spectra of biological systems is to propose a model on a molecular scale and calculate spectra for the molecular changes to see if they are consistent with any portion of the spectra observed in actual samples. This approach has been traditionally a difficult one to follow in the near-infrared because the signals are combinations and overtones of the molecular motions involved. However, continuing advances in instrument and computer technology are making the modeling approach attractive again.
Using a parallel supercomputer, semiempirical and ab initio molecular orbital calculations can be made from a proposed reaction model. Correlation energy calculations using Moller-Plesset perturbation theory and configuration interaction (CI) are typically required to achieve acceptable energy accuracy for fundamental molecular motions. Computation of force constants and harmonic vibrational analysis and determination of intensities for vibrational transitions must be done at the MP2 and CI levels. The calculated spectral model is reconciled to the spectral factor analysis through constrained optimization that uses linear programming to solve the assignment problem, and QQ analysis of residuals to detect model and/or assignment failures. This form of near-IR spectral modeling is being applied to polymers for controlled drug release microspheres, hard gelatin capsules, and lipoproteins in human carotid artery plaque. The figure below shows an intermediate (from lysine side chains) in the crosslinking process of proteins in gelatin for capsules. Over a dozen such intermediates, reactants, and products have been used in the modeling process described above to calculate near-IR spectra for comparison with collected spectra and reconstructed spectra from factor analysis.
Return to the ASRG Home Page.