Sequence Prediction Algorithms vs. Infrared Spectroscopy for Proteins Secondary Structure Optimization

Abstract

More than 80% of the known protein structures are determined using the Macromolecular Crystallography (MX) technique that is feasible in only Synchrotron light facilities. Two difficulties may arise in this case: (1) the easiness of having an access to synchrotrons MX beamlines, (2) the complicated samples preparation of the proteins and large macromolecules, namely the crystallization process. Since the protein structure determination is a long and complicated one, another approach is mandatory to help solving structures in a much faster approach but most importantly a very reliable way. Using computational algorithms coupled with practically simple, faster and feasible experimental techniques to double check the ultimate result proved to be very efficient in the last years. This study seeks establishing a comparative and/or complementary approach of Infrared Microspectroscopy together with computational methods to shorten and to simplify the existing complicated procedures. The study among others, aims at finding a possible solution to overcome the samples preparation phase hence Infrared spectroscopy techniques are requiring almost no prior sample preparation. Proteins secondary structures are the most critical components for their proper functionality. FTIR spectroscopic measurements provide the possibility to obtain information on the secondary structures in their different states. On the other hand, predicting proteins structure can be accomplished through structure prediction algorithms relying on the deposited the proteins sequences in different databases such as the Protein Data Bank (PDB). Information can be obtained by identifying a homolog (homology modeling) of already known structures. The same approach can be followed using ab-initio structure prediction method benefiting from the available structure data (X-ray crystallographic data), although being less informative than the homology modeling ones. We will use a few selected open-source tools, datasets and experimental results to illustrate how effectively the computational methods can serve as a reliable and powerful backup to the commonly used experimental approaches.

The source of the dataset is PDB portal:

http://www.rcsb.org/