PREFMoDeL: A Systematic Review and Proposed Taxonomy of Biomolecular Features for Deep Learning

Of fundamental importance in biochemical and biomedical research is understanding a molecule's biological properties—its structure, its function(s), and its activity(ies). To this end, computational methods in Artificial Intelligence, in particular Deep Learning (DL), have been applied to further biomolecular understanding—from analysis and prediction of protein–protein and protein–ligand interactions to drug discovery and design. While choosing the most appropriate DL architecture is vitally important to accurately model the task at hand, equally important is choosing the features used as input to represent molecular properties in these DL models. Through hypothesis testing, bioinformaticians have created thousands of engineered features for biomolecules such as proteins and their ligands. Herein we present an organizational taxonomy for biomolecular features extracted from 808 articles from across the scientific literature. This objective view of biomolecular features can reduce various forms of experimental and/or investigator bias and additionally facilitate feature selection in biomolecular analysis and design tasks. The resulting dataset contains 1360 nondeduplicated features, and a sample of these features were classified by their properties, clustered, and used to suggest new features. The complete feature dataset (the Public Repository of Engineered Features for Molecular Deep Learning, PREFMoDeL) is released for collaborative sourcing on the web.

Distinguishing features of fold-switching proteins.

Though many folded proteins assume one stable structure that performs one function, a small-but-increasing number remodel their secondary and tertiary structures and change their functions in response to cellular stimuli. These fold-switching proteins regulate biological processes and are associated with autoimmune dysfunction, severe acute respiratory syndrome coronavirus-2 infection, and more. Despite their biological importance, it is difficult to computationally predict fold switching. With the aim of advancing computational prediction and experimental characterization of fold switchers, this review discusses several features that distinguish fold-switching proteins from their single-fold and intrinsically disordered counterparts. First, the isolated structures of fold switchers are less stable and more heterogeneous than single folders but more stable and less heterogeneous than intrinsically disordered proteins (IDPs). Second, the sequences of single fold, fold switching, and intrinsically disordered proteins can evolve at distinct rates. Third, proteins from these three classes are best predicted using different computational techniques. Finally, late-breaking results suggest that single folders, fold switchers, and IDPs have distinct patterns of residue-residue coevolution. The review closes by discussing high-throughput and medium-throughput experimental approaches that might be used to identify new fold-switching proteins.

Methods and applications of machine learning in structure-based drug discovery

Structure-based drug discovery (SBDD) utilizes the three-dimensional (3D) structure of a target protein to identify the lead compounds. This medium is then considered a viable solution based on its availability and correlation with a particular disease. In the case of pandemics like COVID 19, shortening drug development time can save millions of people worldwide;for such a task, classical drug discovery methods will take a long time. Hence, researchers worldwide actively incorporated machine learning (ML) into the drug discovery process, particularly in SBDD, to minimize the lead optimization time. ML uses statistical methods to make a computer perform tasks, take a critical decision, and automate this entire process without being explicitly programmed. With this, the computer can discover new insights about data and unknown patterns crucial to decide the therapeutic use of lead compounds as drugs. The use of ML in the drug discovery field is not new, and it spans an ample research space. By integrating artificial intelligence with ML techniques, viable targets can be found using data clustering, regression, and classification from vast omics databases and sources. In this chapter, we will discuss the methods and applications of ML in SBDD. © 2022 Elsevier Inc. All rights reserved.

Computational Design of Potential Binder Protein for SARS-CoV-2 Spike RBD through A Novel Deep Neural Network Based-Protein Outpainting Algorithm

COVID-19 caused by SARS-CoV-2 is seriously endangering the health of all human beings. There is an urgent need for drugs that can inhibit the replication and propagation of the virus. Traditional macromolecular drugs have long discovery and development cycles and high experimental costs, which can't give rapid response to new viruses. Through computational protein design method, scientists have designed binder proteins with high affinity for the RBD of SARS-CoV-2 spike protein which can effectively inhibit virus replication. However, traditional computational protein design methods rely heavily on human experience and domain knowledge of protein design, and the protein design workflow is too complicated to be widely accepted and used in academia and industry. Based on previous work in the field of deep neural network protein structure prediction and protein design, we developed a novel protein outpainting method that can generate the remaining part of the protein based on a given hot spot motif and complete the entire protein. This method can generate stable protein scaffold which can support the functional hot spot motif, resulting in a protein with excellent thermal stability and developability. We tested this method in a drug discovery project with the aim of designing new SARS-CoV-2 inhibitors. Several proteins are obtained which are predicted to be stable and may have high affinity for the RBD of the SARS-CoV-2 spike protein. Although they have not been verified by wet-lab experiments, we believe that these proteins have great potential to be developed into effective drugs for the treatment of COVID-19. The protein outpainting algorithm proposed in this paper has great advantages over traditional protein design methods. It can be applied to many fields that require the design of functional proteins, such as protein drug design, enzyme de novo design, vaccine design, etc. The method will play an important role in reducing the cost of experiments, shortening the research and development period, and improving the successful rate of biological research and development. © 2021 ACM.

Conformational variability of loops in the SARS-CoV-2 spike protein.

The SARS-CoV-2 spike (S) protein facilitates viral infection, and has been the focus of many structure determination efforts. Its flexible loop regions are known to be involved in protein binding and may adopt multiple conformations. This article identifies the S protein loops and studies their conformational variability based on the available Protein Data Bank structures. While most loops had essentially one stable conformation, 17 of 44 loop regions were observed to be structurally variable with multiple substantively distinct conformations based on a cluster analysis. Loop modeling methods were then applied to the S protein loop targets, and the prediction accuracies discussed in relation to the characteristics of the conformational clusters identified. Loops with multiple conformations were found to be challenging to model based on a single structural template.

Modeling SARS-CoV-2 proteins in the CASP-commons experiment.

Critical Assessment of Structure Prediction (CASP) is an organization aimed at advancing the state of the art in computing protein structure from sequence. In the spring of 2020, CASP launched a community project to compute the structures of the most structurally challenging proteins coded for in the SARS-CoV-2 genome. Forty-seven research groups submitted over 3000 three-dimensional models and 700 sets of accuracy estimates on 10 proteins. The resulting models were released to the public. CASP community members also worked together to provide estimates of local and global accuracy and identify structure-based domain boundaries for some proteins. Subsequently, two of these structures (ORF3a and ORF8) have been solved experimentally, allowing assessment of both model quality and the accuracy estimates. Models from the AlphaFold2 group were found to have good agreement with the experimental structures, with main chain GDT_TS accuracy scores ranging from 63 (a correct topology) to 87 (competitive with experiment).

Probing the Increased Virulence of Severe Acute Respiratory Syndrome Coronavirus 2 B.1.617 (Indian Variant) From Predicted Spike Protein Structure.

Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has led to an outbreak of a pandemic worldwide. The spike (S) protein of SARS-CoV-2, which plays a key role in the receptor recognition and cell membrane fusion process, is composed of two subunits, S1 and S2. The S1 subunit contains a receptor-binding domain that recognizes and binds to the host receptor angiotensin-converting enzyme 2 (ACE2), while the S2 subunit mediates viral cell membrane fusion with the cell membrane and subsequent entry into cells. Mutations in the spike protein (S) are of particular interest due to their potential for reduced susceptibility to neutralizing antibodies or increasing the viral transmissibility and infectivity. Recently, many mutations in the spike protein released new variants, including the Delta and Kappa ones (known as the Indian variants). The variants Delta and Kappa are now of most recent concern because of their well-increased infectivity, both a spin-off of the B.1.617 lineage, which was first identified in India in October 2020. This study employed homology modeling to probe the potential structural effects of the mutations. It was found that the mutations, Leu452Arg, Thr478Lys, and Glu484Gln in the spike protein increase the affinity for the hACE2 receptor, which explains the greater infectivity of the SARS-Cov-2 B.1.617 (Indian Variant).