PPI Prediction - BiopharmaDirect

PPI Prediction

Cells are mainly composed of proteins, and almost every primary cellular process is completed by polyprotein complexes. By identifying and analyzing the components of protein complexes, we can better understand how protein collections are organized into functional units . Since protein-protein interactions (PPIs) are essential for most cell functions, they must be determined to decipher cell behavior. In the past few decades, large-scale PPI analysis has been achieved through technologies such as the yeast two-hybrid (Y2H) system, mass spectrometry and protein chip. However, these methods are time-consuming and expensive, and large-scale experiments usually have a high false positive rate. Computational-based technology to predict protein-protein interactions, which can identify potential PPIs that cannot be discovered by high-throughput methods.

Overall solutions

The study of protein interaction and interaction has very important biological significance. Cells receive exogenous or endogenous signals, and regulate the expression of their genes through their unique signal pathways to maintain their biological characteristics. In this process, protein occupies a very important position, it can internally regulate and mediate many biological activities of cells. Although some proteins can function in the form of monomers, most proteins function with chaperone molecules or form complexes with other proteins. Therefore, in order to better understand the biological activities of cells, the functions of protein monomers and complexes must be well understood, which will involve the study of protein interactions.

Rosetta stone (gene fusion) method

The Rosetta Stone method is based on the assumption that interacting proteins sometimes fuse into a single protein. For example, two or more isolated proteins in a genome can be identified as fused to a protein in another genome. The separated proteins may interact and therefore may be functionally related. In order to identify these sequences, a sequence similarity algorithm is required, such as the algorithm used by BLAST.

Figure 1. Rosetta stone (gene fusion) method.

• AI-based approaches

For protein-protein interaction prediction, artificial intelligence is a more effective method. This AI-based analysis method can also be used for deep neural networks for structure-based drug discovery applications. In summary, AI-based prediction methods can more economically evaluate difficult, tricky and atypical targets, such as protein interactions.

• Conserved gene neighborhood method

The conservative neighborhood approach is based on the assumption that if the genes encoding two proteins are chromosomal neighbors in many genomes, they may be functionally related. This method is most effective in prokaryotes with operons, because the gene organization in the operons is usually related to function.

Figure 2. Conserved gene neighborhood method.

• Classification method

Classification methods use data to train programs to distinguish between positive instances of interacting protein/domain pairs and negative instances of non-interacting pairs. Commonly used classifiers are Random Forest Decision (RFD) and Support Vector Machines. RFD produces results based on the domain composition of interacting and non-interacting protein pairs.

• Association methods

Association methods look for characteristic sequences or motifs that can help distinguish between interacting and non-interacting pairs. A classifier is trained by looking for sequence-signature pairs where one protein contains one sequence-signature, and its interacting partner contains another sequence-signature. They look specifically for sequence-signatures that are found together more often than by chance.

Relationship to docking methods

The field of protein–protein interaction prediction is closely related to the field of protein–protein docking, which attempts to use geometric and steric considerations to fit two proteins of known structure into a bound complex. This is a useful mode of inquiry in cases where both proteins in the pair have known structures and are known (or at least strongly suspected) to interact, but since so many proteins do not have experimentally determined structures, sequence-based interaction prediction methods are especially useful in conjunction with experimental studies of an organism's interactome.

Figure 3. Molecular docking.

Application

The FBDD method has greatly accelerated the identification and discovery of emerging compounds. The most famous example is the discovery and design of vemurafenib, which was successfully marketed. Vemurafenib is the first successful BRAF-V600E kinase inhibitor developed by the FBDD method and approved for marketing for the treatment of melanoma. The first fragments found through the screening of the fragment library are for another kinase, serine/threonine-protein kinase PIM1.

Through a series of optimizations, the vemurafenib molecule was finally obtained. From the structural point of view, Vemurafenib is very stable in the BRAF-V600E active pocket, and the four hydrogen bond interactions formed well fix the binding of Vemurafenib in the pocket. This successful case only took 6 years from the initial selection of fragments to the approval of vemurafenib by the FDA.