Mrs/Miss Ahmad,

The validation of 3D structural models can be carried out correlating between the theoretical binding free Gibbs energy data (deltaGbind) on interacting ensembles of molecules obtained via MM/MD computations with corresponding experimental inhibitory constant (Ki) data on the same ensembles measured in vivo.

Consider, for example, the content of our most recently published chapter of book [1] and the attachment, herein.

[1] Bojidarka Ivanova and Michael Spiteller, Chapter 1. Acaricidal Neuronal Activity from the Perspective of Molecular Level Interactions with Receptor-Proteins: A Theoretical Study; in BOOK: Agricultural Research Updates. Volume 38, Prathamesh Gorawala and Srushti Mandhatri (Eds.) 2021, ISBN: 978-1-68507-182-0; Nova Science Publishers, New York; [https://novapublishers.com/shop/agricultural-research-updates-volume-38/].

Dear Dr Bojidarka B. Ivanova

Thank you very much for your response and sharing..

I really appreciate it

It is not compulsory as a lot of benchmarking has already been done during the development of the programs and the forcefields. Rather it would be better to correlate the simulation results with any available experimental observations. For example, circular dichroism spectra would give you structural information of the DNA which you can correlate with the simulation results.

In simulating DNA, you should be extra cautious about how the base-pair geometries are changing. Base fraying, elongation are very common problems with DNA.

Dear DR Uttam Pal

Thank you for clear explanation.

Dr, based on your experience, what would u suggest the practical way to check in silico work eg. MD simulation, whether its is working or not. I'm just wondering, do we need to repeat the methodology in MD simulation from published article, for example; using same parameter file in gromacs2020 system, and later compare the output? Perhaps, the output should be same as they use same system (gromacs2020). Kindly correct me if I'm wrong.

Really appreciate your recommendation and reponse. Tq Dr

Method validation is a good practice. It is like a positive control and any control experiment is better than no control. However, the conclusion from a positive control can not always be extended to the test system under investigation. Here is an example with molecular docking. In molecular docking, you can take a known ligand-bound crystal structure and then do a self-docking to check if the docking method can reproduce the binding pose of the ligand as observed in the crystal structure (within a reasonable limit of deviation). If the docking method can reproduce the binding pose of the ligand as observed in the crystal structure, you can say that the method is valid and is working for your system. Then you take an unknown ligand for docking and follow the same protocol and find a binding pose with a comparable or maybe even better binding energy. But this does not guarantee that the result is not a false positive. Afterall, it is just a simulation. The only way to validate the docking result is to show at least the binding is actually there. A simple spectroscopy or calorimetry can confirm if there is actually any binding. When there is this experimental confirmation, you can conclude from the in silico result that the simulation did not produce a false positive; the binding is there and the computed binding pose is the best possible binding pose. The same goes for the MD simulation. Now coming to your question, if you try to reproduce a published result keeping all the conditions similar (software, parameters etc.), you should get a similar result. But that is not method validation. Method validation should also be tied to experimental observation. Say you take a solution NMR structure of a DNA fragment from a nucleotide database and then run MD following the protocol that you want to use for your experiment and see if the structure remains intact. If the equilibrium structure remains the same as the input, you can say that the method worked for such a system (a known DNA fragment from database) and if your system is similar (e.g. a DNA fragment with different length and composition), it probably should work for your system as well.

Dr Uttam Pal

Thank you very much for your crystal clear explanation.. I really appreciate it