Computational Molecular Biology 2024, Vol.14, No.3, 115-124 http://bioscipublisher.com/index.php/cmb 117 binding sites and docking solutions (Lima et al., 2016). The complementary use of SBDD and LBDD, along with their integration with experimental data, has been shown to enhance the efficiency of drug discovery processes (Macalino et al., 2015). 3.3 ADMET prediction and optimization The prediction and optimization of absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties are critical for the development of safe and effective drugs. Computational tools have been developed to automate the evaluation of ADMET properties, reducing the time and cost associated with drug discovery (Rosales-Hernández and Correa-Basurto, 2015). Recent advances in computational chemistry have enabled the accurate prediction of ADMET profiles, which is essential for optimizing lead compounds and designing novel biologically active molecules. The integration of machine learning techniques with traditional computational methods has further improved the prediction accuracy of ADMET properties, facilitating the development of drugs with favorable physiological profiles. 4 Challenges in Computational Chemistry 4.1 Accuracy and reliability of predictions One of the primary challenges in computational chemistry is achieving high accuracy and reliability in predictions. Despite significant advancements in computational methods, accurately predicting ligand binding affinities remains difficult. For instance, free-energy calculations, which are crucial for predicting binding affinities, have seen improvements in force fields and sampling algorithms. However, achieving the necessary accuracy to guide lead optimization reliably is still challenging, limiting their widespread commercial application (Wang et al., 2015). Additionally, while methods like FEP+ (Free Energy Perturbation) have shown promise in predicting protein-ligand binding free energies with high accuracy, there are still limitations in their implementation that need to be addressed to ensure consistent reliability across different targets and ligands. 4.2 Computational costs and resources The computational costs and resources required for advanced simulations pose another significant challenge. High-accuracy methods such as free-energy calculations and molecular dynamics simulations are computationally intensive, often requiring substantial hardware resources and time. The need for extensive computational power can be a barrier, especially for smaller research groups or institutions with limited access to high-performance computing facilities (Abel et al., 2017). Moreover, the complexity of the systems being simulated, such as large protein-ligand complexes, further exacerbates the demand for computational resources (Decherchi and Cavalli, 2020). Despite the advent of low-cost parallel computing, the resource-intensive nature of these simulations remains a critical challenge. 4.3 Integration with experimental data Integrating computational predictions with experimental data is essential for validating and refining computational models, yet it presents its own set of challenges. Computational methods must be iteratively validated and adjusted based on experimental results to ensure their accuracy and applicability. This iterative process can be resource-intensive and time-consuming (Persico et al., 2016). Additionally, the complexity of biological systems often requires the use of multiple computational tools and methods, each with its own set of parameters and assumptions, making the integration process even more challenging (Rosales-Hernández and Correa-Basurto, 2015). Effective integration also necessitates multidisciplinary collaboration, combining expertise from computational chemistry, bioinformatics, and experimental biology to achieve meaningful and actionable insights. 5 Advances in Computational Methods 5.1 Machine learning and ai in drug discovery 5.1.1 Applications of AI in drug design Artificial intelligence (AI) has significantly transformed the landscape of drug discovery by enabling the efficient identification of new chemical entities with desirable properties. AI algorithms, particularly deep learning, have been applied to various stages of drug discovery, including structure- and ligand-based virtual screening, de novo
RkJQdWJsaXNoZXIy MjQ4ODYzNA==