China
Africa
Explore chapters and articles related to this topic
Role of AI in the Advancement of Drug Discovery and Development
Published in Utpal Chakraborty, Amit Banerjee, Jayanta Kumar Saha, Niloy Sarkar, Chinmay Chakraborty, Artificial Intelligence and the Fourth Industrial Revolution, 2022
Shantanu K. Yadav, Poonam Jindal, Rakesh K. Sindhu
In many biological procedures, protein-protein interactions (PPIs) are not crucial but they are linked to several diseases (Scott et al., 2016). The PPI architecture often contains concentrations of protein-protein binding locations, which represents a fresh goal (Santos et al., 2017) that really varies from conventional objectives (ion channels [G-protein], kinases, nuclear receptors) to extend the destination space as well as encourage the growth of tiny molecular medicines. Thus, in addition to the introduction of protein annotation, an in-depth learning of the interface zone of PPIs is important for drug design centered on complicated protein-protein construction and the treatment of associated ailments (Wilson et al., 2013). Many PPI prediction calculation methods have been generated. Due to drawbacks of modern PPI approaches, such as high price, long-term commitments, large quantities of data noise, and strong false positive and negative levels, PPI knowledge is very limited and restricted. The current PPI prediction, therefore, covers mostly two structure- and sequence-based categories. In particular, the method based on the protein prototype structure is more reliable and simpler as most PPI interfaces are conservative. For example, Maheshwari and Brylinski (2016) developed a template structure-based eFindsite PPI prediction approach to identify residues of PPIs from a weak homologous template. This approach is predictively accurate both in the experimental protein configuration and in the protein, structure created by silicon. The PPI interface can be modeled by using the protein-protein docking model on the basis of the complementary theory when the 3D configurations of the two interacting proteins are identified (Vakser, 2014). Structure- based prediction algorithms conduct better than sequence-based techniques, even with limited quantities and performance for recognized protein structures. For example, presently known bacteria, yeast, or humans contain little organizational data for 80 percent of PPIs. AI has made important strides in anticipating PPIs using sequence-based techniques, with exponential growth in protein sequence information (Mosca et al., 2013). Du et al. (2016) used integrated hidden Markov models (ipHMMs) in 2016 to obtain Fisher fractional characteristics from protein sequences. A stacked autoencoder was used to build a DNN perfect for estimating protein residues in an interaction matrix. The DL model’s general forecast precision is 80.82%, that is 15% greater than that of the old-style ML model.
Evaluation of a single amino acid substitution at position 79 of human IFN-α2b in interferon-receptor assembly and activity
Published in Preparative Biochemistry and Biotechnology, 2019
Samira Talebi, Alireza Saeedinia, Mehdi Zeinoddini, Fathollah Ahmadpour, Majid Sadeghizadeh
The interacting residues of IFN-α2b and IFNAR1 were known using literature reviews. Molegro Virtual Docker was applied to visualize these selected residues and they were confirmed by biological perspective. Protein–protein docking for both native and mutant forms was prepared through HADDOCK 2.2 web server (http://milou.science.uu.nl/services/HADDOCK2.2/haddock.php). In the HADDOCK web server, to sum up, the final score, the weighted sum of the parameters including electrostatic energy, desolvation energy, binding energy, distance restraint energy, van der Waals energy, and buried surface region is used. The concluding score of the HADDOCK web server explains two proteins binding power.