Our expertise
Biomolecular modelling encompasses a diverse set of in silico methodologies used across multiple areas of biological and chemical research. Beyond general biomolecular modelling - which includes the analysis of biomolecular structure, function, and interactions - our research group is particularly focused on the discovery and evaluation of bioactive compounds. In addition to newly synthesized molecules, we are dedicated to investigating naturally occurring, food-derived bioactive compounds (such as bioactive peptides, polyphenols, and related metabolite classes) that have the potential to modulate physiological processes and thereby contribute to the improvement of human health. The core areas of expertise within our research team are outlined below.
Lead discovery, optimization and rational drug design
We conduct the essential in silico stages required for successful drug design, including target identification and validation, lead discovery and optimization, as well as preclinical computational evaluation. Our molecular modelling approaches are grounded in current scientific knowledge and methodological expertise, ensuring reliable and accurate results.
Computer Aided Drug Discovery
Computer-Aided Drug Discovery (CADD) employs advanced molecular modelling methodologies to identify and evaluate potential drug targets and to optimize candidate molecules. CADD enables the identification of small molecules that interact with and bind to specific biological targets, as well as the prediction of the safety and efficacy of drug candidates through comprehensive in silico simulations. Our team applies both ligand-based and structure-based drug discovery approaches.
1.Ligand based drug discovery
-Pharmacophores
-SAR/QSAR/3D-QSAR
2.Structure based drug discovery
-Virtual screening
-Molecular docking
-Molecular Dynamics
-MMGBSA
-Structure based de novo drug design
-Structure based optimization
Rational drug design
By integrating insights from both ligand-based and structure-based drug discovery approaches, along with molecular dynamics simulations, we perform lead optimization and rational drug design.
Target identification of newly synthesized organic compounds
Newly synthesized organic compounds often lack information regarding their potential biological activity. We employ a wide range of in silico techniques to predict likely molecular targets for these small molecules. Once experimentally validated, the resulting insights can reveal new biological roles and applications for the compounds.
Using in silico techniques we are able to make a small selection of potential inhibitors that should be further tested using in vitro methods.
Designing of proteins in order to provide desirable properties and reactions
Protein engineering – including the modification or design of novel proteins and enzymes through point mutations or the construction of chimeric variants – represents a powerful strategy for developing new or enhanced biotechnological solutions. For example, improving thermal stability is a critical requirement in the development of commercially viable enzymes used in industrial process engineering. To support such advancements, we apply cluster analysis and ancestral sequence reconstruction to identify suitable sites for targeted mutations.
Wandi, et al., 2020
doi:10.1111/febs.15192
Investigation of enzyme reactions
Comprehensive analysis of potential energy of chemical reactions and transition states are crucial steps in their optimization and design of new reactions
Potential energy of chemical reaction
Potential energy surface
Time-dependent Shcrödinger equation
Interaction of biomolecules with materials
Understanding how biomolecules interact with artificial materials is of critical importance across a broad range of biotechnological applications. Such interactions play a key role in the design and functionalization of biosensors, the development of drug delivery systems, and the engineering of materials for food packaging and preservation.
By studying these interfaces at the molecular level, we aim to optimize binding properties, stability, biocompatibility, and functional performance. Our research contributes to the rational design of advanced biomaterial systems that support improved efficiency, safety, and sustainability in applied biotechnology.
Literature research
A thorough and comprehensive review of the literature is the first and most essential step in any scientific investigation. Detailed knowledge of the subject area and awareness of the current state of research are critical for formulating a well-founded hypothesis. Up-to-date data must be carefully examined to ensure accuracy and to avoid conclusions based on outdated or misleading information.
Our team conducts professional, systematic literature reviews by examining reputable scientific databases and leading international journals (including PubMed, ScienceDirect, and others). This ensures that our research is grounded in the most current and reliable scientific evidence.
Database search
We provide systematic database searches and data collection services relevant to the research topic of interest. This includes mining information from major biological and chemical databases such as NCBI, PDB, UniProt, ZINC, BRENDA, DrugBank, and others.
Accurate data on the availability and characteristics of RNA/DNA and protein sequences, 3D structures, ligands, and associated metadata (such as annotation quality, resolution, and experimental validity) are essential for ensuring the reliability and success of any research project. We evaluate the suitability and usability of these data resources to support informed decision-making in downstream computational and experimental workflows.
Sequence Analysis
Analysis of RNA/DNA and protein sequences using reliable and proven software. In order to analyze sequences we perform multiple sequence alignment, sequence annotation, cluster analysis, search for individual conserved nucleotides and residues, construct DNA primers, etc. Molecular evolution is a basis for diversity of life. Therefore, sequence analysis provides us valuable information regarding relations among sequences and changes with functional consequences. For example, sequence analysis can provides us information regarding individual responses to diseases and drugs, which is one of the fundamental cores of personalized medicine.
Vukic, et al., 2015
DOI: 10.1080/08905436.2015.1059766
Analysis of compounds structure
Detailed and comprehensive analysis of available structural data is essential for understanding mechanisms of compounds` activity as well as for development of a strategy for activity modification (inhibition or enhancement). Activities of small molecules can be efficiently modified through in silico chemoinformatic analysis which falls upon our field of expertise. High quality structure of target molecule is highly important for drug design. Analysis of active site as well as allosteric binding sites and understanding of their properties provides framework for rational drug design through modifications of interactions with the target molecule (covalent, electrostatic, hydrophobic, water bridges, etc.).
Wandi, et al., 2020
doi:10.1111/febs.15192
Wandi, et al., 2020
doi:10.1111/febs.15192
Maruf, et al., 2006
doi:10.1038/nature04716
Homology modelling
Based on BLAST search for homology sequences we construct (modeller) and evaluate (Rhamachandran plot, Qmean analysis, visual inspection) homology models of proteins.
Vukic, et al., 2015
DOI: 10.1080/08905436.2015.1059766
Virtual screening
One of the first steps in computational search for new drugs is virtual screening. We are able to screen large offline databases of small molecules that are stored on our local machines (FIMM database (140 000 compounds), UNPD - Universal Natural Products Database (220 000 compounds), TCM Database@Taiwan (20 000 compounds)). Natural products are desirable for screening due to wide diversity of compounds as well as good purchasability. Furthermore, we are able to construct desirable database for virtual screening using ENAMINE, ZINC, MolPort, ChEMBL or other online databases. We screen molecules using several methods:
-Molecular docking simulation:
– High throughput virtual screening
-Pharmacophore based screening:
- based on protein ligand complex
- based on receptor cavities
- based on receptor residues
- based on multiple/single ligands
Project in progress
Molecular docking simulations
High precision molecular docking simulations (covalent and non-covalent) are essential in prediction of ligand binding to the target molecule in absence of experimental data. In order to select a potent inhibitor, accurate prediction of ligand`s potential orientation is crucial for determination of binding energy. Several techniques and software have been shown to be very effective in prediction of protein-ligand complex. For non-covalent docking we prefer Glide and Autodock due to high precision, flexible residues and possibilities of manual docking, while we use CovDock for covalent docking simulations.
Calculation of binding energies
Although calculation of absolute or relative binding energies using in silico techniques is still a highly challenging process, some progress have been achieved in recent years. Using high end techniques and software (MMGBSA, FEP+) we are able to narrow selection of potential inhibitors which should be further tested using molecular dynamics simulations and in vitro techniques.
Molecular Dynamics simulations
Information about stability and movements of modelled structures and complexes are highly important to understand dynamics of the subject of interest (or process). We use Desmond to setup the system and perform MD simulationst on in house and distant servers equipped with supercomputers (CPU and GPU).
Wandi, et al., 2020
doi:10.1111/febs.15192
SAR /QSAR / 3D-QSAR
Correlation between structure and activity/property is a basis for ligand based drug design. The prediction of activity based on molecular structure is a very important step in synthesis of compounds with desired biological activity and rely on assumption that structurally similar molecules have similar biological activity. In order to predict compounds properties we construct different models (selective QSAR, CoMFA, CoMSIA, etc.) and use different regression and validation methods (Artificial neural networks, R2, Q2, etc.)
Vukic, et al., 2022
doi: 10.1007/s11094-022-02686-z
Vukic, et al., 2017
doi: 10.1016/j.nutres.2017.07.009
Ligand based pharmacophores
We search for common chemical features of one or more (aligned) ligands that contribute to the binding interactions at the target protein and are essential the biological activity:
- hydrogen bond acceptors
- hydrogen bond donors
- hydrophobic groups
- aromatic groups
- positively charged group
- negatively charged group
Using pharmacophores we:
- predict ADMET properties
- explain the structure-activity relationships of a series of ligands
- search for novel active ligands through virtual screening of compound libraries
Protein – peptide, protein – protein docking
We perform protein – peptide and protein – protein docking and analyze their interactions. Peptides are one of the main signaling molecules and modulators in human body which put them in the focus of scientific research in recent years. Protein-protein docking aims to predict the structure of a complex formed between two or more protein molecules. These interactions are fundamental to many biological processes, including enzyme-substrate recognition, antibody-antigen binding, and signal transduction pathways. By understanding how proteins interact, we can uncover the mechanisms of diseases, identify potential drug targets, and design molecules that can interfere with or enhance these interactions.
Project in progress
Nucleic acids interactions
We investigate interactions of nucleic acids (both DNA and RNA) with proteins, which give us information regarding transcription, translation, regulatory or some other process, as well as interactions with small molecules.
Project in progress
Jaric, et al., 2023
ADME-Tox preclinical investigation of potential drugs
Pharmacokinetic properties of a drug candidate is extremely important for its further development as a drug. In order to safely and efficiently act, the drug must have adequate ADME-Tox properties (Absorption, Distribution, Metabolism, Excretion and Toxicity). We use various in sillico tool in order to predict ADME-Tox properties of desired compounds.
Data analysis, machine learning and artificial neural networks
Our team performs data analysis (DA), machine learning (ML), artificial neural networks (ANNs) and other statistical methods in order to achieve our goals. DA, ML and ANN are revolutionizing biotechnology, medicine, and drug design by enabling unprecedented advancements in these fields. ANNs, inspired by the human brain's architecture, are powerful tools for modeling complex biological processes and patterns within large datasets. By leveraging DA techniques, researchers can extract meaningful insights from vast amounts of biological data, such as genomic sequences, protein structures, and patient health records. ML algorithms further enhance this process by identifying patterns, predicting outcomes, and optimizing experimental designs. In medicine, these technologies facilitate early disease detection, personalized treatment plans, and the development of predictive models for patient outcomes. In drug design, ML accelerates the discovery of new therapeutics by predicting molecular interactions and optimizing drug candidates. Together, these technologies are driving significant progress in understanding diseases, developing innovative treatments, and improving patient care.
