17th Global Drug Delivery & Formulation Summit

Molecular Modeling and Machine Learning for Small Molecule and Biologic Drug Formulation

Speaker:
John Shelley, Fellow at Schrödinger

Abstract:
Selecting and combining the right ingredients in the appropriate manner is essential for successful drug formulation given the inherent challenges and competitive market. With advances in modern machine learning, physics-based simulation techniques and computer hardware, modelling is emerging as a valuable source of information that complements experimental characterization.  We showcase a cross-section of capabilities within Schrödinger’s Suite for modeling related to formulations of small-molecule or biologic drugs.  For small-molecule drugs workflows have been created for characterizing crystal polymorphs, crystal morphology and degradation risks as well as calculating elastic constants (bulk modulus, shear modulus, etc.), powder diffraction patterns, glass transition temperatures (Tg), diffusion constants, pKa values, melting points, water adsorption and various solubilities. For biologics our toolset supports homology modeling, and the calculation of aggregation propensity, titration curves, isoelectric points and viscosity among other things.  Complex and evolving structures, often in fluid states, play a crucial role in the pharmaceutical industry.   For both small-molecule and biologics formulations powerful simulation tools employing atomistic or coarse-grained models to permit the characterization of molecular interactions and nanoscale structuring, sometimes within otherwise disordered bulk systems (e.g., LNP formation, self-assembly of polymer-based structures, dissolving amorphous solid dispersions, liposomes and protein-excipient interactions).