Let's talk
Contact us

Q&A: Formulation Development for Small Molecules and Biologics Enhancements with Molecular Modeling

 Dr. John Shelley, Schrodinger

 Dr. Andrea Browning, Schrodinger

Q: Do you offer formulation Analysis? For example, for Cannabis, Delta-8, Delta-9, and THC to be specific.?
A: For this, you likely need experimental characterization which we do not offer as a service.

Q: What’s your ability to predict polymorphs of an API? And what are the conditions that promote those polymorphs?
A: Different crystal structures of the same chemistry can be simulated in the tools discussed in the webinar. For those interested in polymorph prediction, please reach out to us directly for a more detailed discussion. 

Q: Do you use any AI for these simulations?
A: Machine learning and other QSAR approaches are available in our software. The examples shown in this webinar focused on physics based modeling. The addition of Machine learning and QSAR can increase the predictive power of simulation and provide insight in numerous other areas. Our website and Summer of Science Series has more information on these tools.

Q: Which software do you use for simulating the dissolution of amorphus polymers?
A: Most of these were molecular dynamics simulations for which we use, Desmond which runs on a GPGPU

Q: What details of the molecule you should insert to the model?
A: This really depends on the problem. For instance, if you are interested in bond-breaking or bond-forming then you usually need to use a quantum mechanical approach with all of the atoms represented. At the other end of the scale if you are looking at molecular association in solution simplified models such as coarse-grained models can be applicable.

Q: How do you differ between different polymorphism?
A: With the PXRD tools shown, the XRD pattern for different crystal structures can be found. With experimental patterns, differences in the lattice dimensions can also be found. With known crystal structures, the impact of that structure on properties such as the morphology can be explored. 

Q: What are the main deficiencies in Molecular Modeling?
A: There are quite a few, however, molecular modeling is becoming more broadly applicable. An important thing to consider is that often no one technique stands on its own whether it is experimental, theoretical or computations. It is the complimentary use of techniques that often moves projects ahead. Here are some weaknesses for Molecular Modeling:

a. Accuracy of the model to represent the molecules of interest and their interactions
This depends on the general approach used (QM, all-atom or coarse-grained). These days QM and all-atom models for organic molecules have become very reliable. The validated coverage for coarse-grained models is an important thing to consider.

b. For some types of problems molecule conformations and higher-level molecular structures need to form or evolve. Some important processes are slow and thus require a lot of computer time.

c. Often one calculation on its own is not sufficient. Instead multiple calculations and usually different types of calculation are needed in a validated workflow to be useful. Workflows do not exist for all types of problems.

Q: What kind of molecular modelling software(s) are available for use as open source?
A: There are quite a few. You should feel free to explore the web to find out more. What we offer as a commercial entity is layers of interfaces, excellent support, and validated workflows to help researchers focus more on the problem rather than how to run the calculations.

Q: Phytosomes have also been in drug delivery system. Which is better?
A: Some lipid-based formulations have complex mixtures of ingredients and thus could encompass phytosomes. In general, what is best is problem-specific… depending on what is the active ingredient(s), how much of the active ingredient is needed, over what time period does the drug need to be delivered and to whom (e.g., adult, child, conscious, unconconscious).

Q: Most Molecular Modeling software even Hansen can only predict positive Chi values, yet Chi can be negative as well. Does your model able to predict negative Chi?
A: Solubility parameters are by definition positive. The chi value can be obtained from energy of mixing definitions and therefore simulations as well and in those cases can be negative or positive. With our software, energy of mixing can be obtained.

Q: On the shear modulus, perhaps I missed the detail, is this for the solid crystal, amorphous form, or do you also look at how different material particles interact when being compacted as well?
A: This calculation can be performed for the solid crystal, amorphous form, and excipients. For larger material particles (eg micron sized), the information from the molecular level simulation can be used as input into FEM simulations.

Q: Do the protein solutions exhibit thixotropic behavior?
A: Protein solutions in general can exhibit this behavior. I suspect that our models would as well however we have not investigated this.

Q: What happens after the drug is released? The leftover liposome?
A: Natural or near natural lipids are efficiently removed from circulation. Some lipids are designed to slow down this process and testing in those cases is likely required to ensure safety.

Q: What do you think about the best: high or low Young modulus?
A: It depends on the material in question and the use case.