Computational Modeling and Simulation in Biomedical Research

The digital revolution has significantly impacted worldwide technologies over the past few decades. Biomedical research is one of the most impacted fields with the advancement of computational power and data processing. The human genome sequencing project has generated an enormous amount of information, which is challenging to be stored and interpreted without the aid of computer programs. The development of computational algorithms has, therefore, greatly eased and reduced the time to study and analyze the human genome information. This has directly improved our understanding of the complex genome structure such as the presence of different regulatory regions and non-coding regions that code RNA like microRNA or long noncoding RNA (lncRNA). In addition, many computational tools have been developed to improve our understanding of the biomedical field. This covers the areas from the study of biomolecule structures and interactions, dynamicity of biomolecules, cellular activity, to system biology. This chapter thus provides a brief introduction to various computational tools in these areas and their importance.

 There has been a great development in the field of computational modeling and simulation in biomedical research during the last ten years, in particular, in brain stimulation of Parkinson’s disease (PD) patients and, recently, even in that of Alzheimer’s disease (AD) patients. Computer modeling allows such electrical stimulations using statistics, bioinformatics and advanced machine-learning algorithms. The current book chapter discusses the advantages of computational modeling in studying biomedical research. Using computational modeling, classification algorithms can be applied to microarray and RNA sequencing data (such as hierarchical clustering - HCL, t-SNE and principal component analysis - PCA), and high-resolution images can be generated based on the analyzed data and patient samples. Additionally, genomic data can be analyzed from cancer patient samples carrying mutations or exhibiting aneuploidy chromosomal changes (such as lung cancer, breast cancer, cervical cancer, ovarian cancer, glioblastoma and colon cancer). Also, microRNAs (miRNAs) and long noncoding RNAs (lncRNAs) can be analyzed. We can identify cellular vulnerabilities associated with aneuploid, and assigned aneuploidy scores can generate mushroom plots on the data. Functional network analyses can highlight altered pathways (such as inflammation and alternative splicing) in patient samples, and cellular composition and lineage-specific analyses can highlight the role of specific cell types (e.g., neurons, microglia – MG oligodendrocytes- OLGs, astrocytes, etc.). Computational platforms/tools, such as Matlab, R, Python, SPSS and MySQL, can be used for analysis. The data can be deposited in the Gene Expression Omnibus (GEO). CRISPR/Cas genomic targets can be identified for therapeutic intervention using computer simulations, and patient survival curves can be computed. Further comparison to mice models can be made. Additionally, human and mouse stem cells can be analyzed, and non-parametric gene ontology (GO) analyses using KolmogorovSmirnov (KS) statistical tests can be applied to microarray or RNA sequencing data.

The super-fast automated next-generation sequencing (NGS) technology allows parallel sequencing of millions of genomics molecules with higher accuracy at low cost and less time consumption. However, the three-dimensional structure needs to be determined for experimental purposes, and solving the three-dimensional structure of a biomolecule via the alternative experimental approaches requires special skills and equipment, which is time-consuming, labor demanding and expensive. This situation resulted in the widening of the space between solved biological molecules and known sequences. Recently, bioinformaticians and computer scientists have developed computer-based algorithms and protocols to solve these issues. The developed computational approaches and algorithms allow researchers to 1) perform prediction and refinement of models close to their native molecule structure based on the data from other molecules possessing similar homology, 2) predict potential interaction and possible reactions between two biomolecules and 3) gain meaningful insight when it is impractical to be obtained via theoretical or experimental analysis. The development of these computational algorithms allows scientists to discover, predict and study important molecules at a faster pace. This chapter will introduce readers to the basic computational algorithms used to develop advanced bioinformatic protocols and tools. 

In all living organisms, proteins carry out essential biological processes. The biological function of a protein depends on the building blocks of amino acids that fold into three-dimensional architecture. Understanding the protein structure-function relationship will, therefore, allow the generation of hypotheses on how to inhibit, control, or modify protein for better use in biomedical research, especially when dealing with emerging infectious diseases. Due to the exponential growth of protein sequence data but not the structural data, protein modeling thus provides an alternative approach to shed some light on the structure of a protein. Protein modeling has the advantage of solving the protein structure as it is relatively faster and cost-effective than experimental means. With the availability of the structural information of a protein, the function of the protein can be further understood. Hence, rational engineering or protein design with improved functionality can be performed and can be useful in biomedical research. This highlights the increasing importance of protein modeling in biomedical studies. This chapter provides a brief overview of existing protein modeling techniques. The applications of protein modeling in recent biomedical research are also summarized here.

Molecular recognition is one of the key principles in the development of active pharmaceutical compounds. Active molecules that can be delivered in vivo to a biological target, responsible for pathological states associated with a disease, can be developed into therapeutic agents. Such molecules must overcome relevant biological barriers and establish intermolecular interactions with the target in order to modulate its activity. The drug discovery process entails the identification of potential therapeutic agents and the design of optimal formulations for targeted or prolonged drug release in vivo. This requires a balanced and dynamic interplay of interactions between the therapeutic agent and different molecular systems through diverse environments. Computational methods, including molecular dynamics simulation, complement experiments in the evaluation of relevant biochemical processes at different stages of drug development, e.g., the elucidation of the ligand mode of action. In this chapter, we will explore the applications of various molecular modeling approaches to evaluate the key interactions small molecules form with different targets. Molecular docking is the most common tool used to evaluate the ligand complementarity to the target binding site. Although the flexible receptor and induced fit approaches provide some additional insights into how target flexibility affects ligand binding, biomolecules have a large number of degrees of freedom, often demanding the use of more exhaustive sampling methods to explore the ligand-binding associated conformational dynamics. This can be achieved with molecular dynamics and enhanced sampling approaches to model large conformational changes. In particular, molecular dynamics of protein-ligand complexes can describe the plasticity of the protein binding sites by identifying dynamic pharmacophores―dynophores. These pharmacophore models incorporate information on target flexibility and describe the dynamics of intermolecular interactions. We will provide a relevant introduction to the above-mentioned techniques and explore key successful applications in hit discovery and lead optimization efforts of drug development campaigns.