In recent years, macrocycles have attracted attention as a promising therapeutic modality for many indications. Macrocycles are larger than traditional "rule of 5" small molecules and can often be more selective, allowing them to target binding sites that are challenging for smaller molecules, like protein–protein interactions. Unlike biologics, though, macrocycles can be engineered to be orally bioavailable and cell-permeable, making them "the best of both worlds."
Despite their promise, designing effective macrocyclic drugs introduces a number of important complications. Macrocycles exhibit very complex conformational behavior: due to the high number of coupled torsions in a macrocyclic ring, even small modifications can lead to vastly different final geometries. This complex landscape leads to activity cliffs and unintuitive SAR, making it challenging to rationally design better binders.
Many macrocycles contain hindered internal torsional degrees of freedom, which can result in the formation of stable atropisomers and additional chiral elements. When present, these atropisomers can dramatically affect binding affinity but often require specialized enantioselective syntheses and analytical techniques. Even determining whether or not a given macrocyclic torsion will result in atropisomers often requires time-consuming spectroscopic experiments.
Developing macrocycles with the right biophysical properties requires careful control of ionization state: many macrocycles need to be protonated to be water-soluble, but too much ionization prevents compounds from getting into cells. The complex network of non-covalent interactions and internal hydrogen bonds often observed in cyclic peptides can have dramatic effects on the pKa of basic or acidic residues, which in turn can cause empirical pKa prediction algorithms to fail.
Rowan makes it fast and easy to run accurate simulations of macrocycles and their properties. Our conformational search workflows use physics-based methods and modern ML techniques to ensure that even complex conformational changes are modeled quickly and accurately. We even suppport the addition of arbitrary geometric constraints to conformational searches, allowing scientists to use experimental NMR data to refine simulated structures.
To determine if atropisomers are present, our software contains pre-defined workflows for quickly generating torsional energy profiles. Rowan makes it possible to use modern machine-learned interatomic potentials to compute torsional barriers at a fraction of the cost of conventional quantum mechanics-based simulations, dramatically accelerating the speed at which these calculations run.
Rowan makes it possible to predict and visualize the effect of structural modifications on macrocycle ionization. Our state-of-the-art physics-based pKa prediction workflow naturally takes into account conformational and through-space effects on basicity and acidity while running orders of magnitude faster than comparable methods.
