Welcome to my Homepage.
Dr Ramin Ekhteiari Salmas, PhD, MRSC
Simulation and Machine Learning:
Molecular modeling approaches can be a powerful way in studying of biological systems in either atomic or molecular scales. These methods provides us to investigate the chemical interactions formed between atoms of proteins, which lead to their secondary and tertiary structures. Classic molecular dynamics (MD) simulations which generate thousands of trajectory frames in a period of time are feasible in sampling of structural conformations and essential in conformational transitions of proteins. Docking simulations which is considered as a low time-consuming method can calculate the energy values released upon protein-protein or ligand-protein bindings. There various methods including Poisson–Boltzmann or generalized Born and surface area continuum solvation (MM/PBSA and MM/GBSA) which are used for free energy calculations of biological systems. Monte Carlo (MC) methods are time-independent simulations sampling different conformers of proteins based on random protocols. Our purpose is to calculate the statistical thermodynamics terms while the dynamics of atoms are being mimicked using different simulation methods or umbrella sampling. In order to investigate the conformational changes of a system which lead to signal transferring, umbrella sampling method is highly recommended. Normal mode analysis (NMA) and principle component analysis (PCA) are the other statistical methods which are implemented to study the dynamics and fluctuations of atoms.
Computer-Aided Drug Design (CADD):
Molecular modeling methods are used in drug discovery studies, in which large ligand data based can be virtually screened and ranked according their binding affinities inside the targets. Different methods are mainly implemented for this purpose, including docking simulations and Quantitative structure–activity relationship models (QSAR models). The ligands reported from these methods can be good candidates for further analysis in vitro and in vivo. Virtual screening of large data base can provide to have more chance in discovering new potent ligand since these screenings are low-time consuming and not expensive as much as experimental methods. It's highly recommended that the binding energy of new compounds inside targets which are being selected for testing in vitro or in vivo, are predicted theoretically before testing them experimentally. It can provide initial information about the ligands' properties at that stage, which can be helpful on how to select samples and which samples should be incorporated in the tests. In short, in silico approaches are believed to be a reasonable complementary to either in vivo or in vitro tests. It must be kept in mind that a good docking score of a compound doesn't mean always that it's a potent ligand for its target. The docking scores are estimated with lots of assumptions which can affect the final results, so they should be considered as initial data which forwarded to the further analysis. No docking score methods can solely judge about either the activity or inactivity of a compound. I'm quite against to the calculated binding energies reported without any experimental approves - these reports can be challenging and may conflict with the experimental data, which will be being reported in future studies about those compounds.
Programming for biologists and medical scientists:
I would so like to begin by writing a quote from Stephen Hawking:
"Whether you want to uncover the secrets of the universe, or you want to pursue a career in the 21st century, basic computer programming is an essential skill to learn."
I highly recommend to the scientists, who work in molecular modeling fields, to be professional at least in one computer programming language - it can be Python, C++, C, TCL or FORTRAN. It must be noticed that we can be smarter that any computers but not faster, so in order to enhance the speed of our calculations, we need to program computers to work for us. I personally prefer to work with an object-oriented language like Python. All the purposes which are required in bioinformatic studies can be easily coded in Python. In addition to the programing by these facilities, we should not forget to develop new modules using our mathematics knowledge. We must keep in our mind that there will be no science, unless mathematics formulas are implemented. Mathematics is considered as a potent key in solving the problems in our researches - it can help us to think how to deal with the problems prior solving them. You are welcome to write me for any help in this area.