Determination of protein sequence VS Determination of novel protein 3D structures

Proteins being the building block of life therefore they can be considered as molecular devices that perform various biological functions. It is this basic structure that allows small molecular interactions to take place in cells. So prediction of structure helps us to determine the function of protein (s) along with its relation with other protein(s). Structure prediction has efficiently increased the process of drug finding in combination with other rational approach(s).

There are 42015 distinct protein sequences (as per PDB statistics), 140090 human sequences (as per UNIPROT statistics) as compared to 37160 3D structures of human sequences.For getting novel 3D structure of protein to join the fast speed of sequence determination, computational methods are reciprocating to the experimental methods. So, using classical biology, mathematical, statistical and computer science structural bioinformatics has evolved that can help us to comprehend, manage and explore the structural information associated to the molecules.

Protein structure prediction through computational methods compensates the limitations of experimental methods. Computational method involves comparative modeling of proteins, commonly known as molecular modeling. These include homology modeling (template-based) and de novo modeling. Homology modeling is more accurate as it is based on basic observation that all the members of protein family shows the same fold (s); also it gives protein (target)model with the available sequence depending on experimentally determined structures of homologous protein. Obviously accuracy is more for close homologous structures but good models can be generated with low sequence similarity. Various online servers and tools are available for homology modelling like SWISS MODEL , RAPTOR X, PHYRE 2, MODELLER, Protein Model Portal etc..

This is followed by model refinement that is removal of clashes, bond lengths’ geometrical regularizations or other structural amendments. The structure thus generated should be validated to check errors and inaccuracies. Most important, the requirement of structure quality depends on the use.

Reference:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4112578/

https://www.ncbi.nlm.nih.gov/books/NBK6824/