Table of Contents
select drug targets that are implicated in diseases that represent large market opportunities where there are significant unmet medical needs and where the target has validated
biology but has been difficult-to-drug. We believe that our proprietary drug discovery platform Contour provides a competitive advantage in solving difficult-to-drug targets, or those that have not
solved using conventional high throughput library screening and optimization. We believe targets with validated biology, achieved through genetics, animal data and direct clinical experience,
increases the probability of success of our discoveries. We frequently rely on mouse gene knock-out models or human genetic diseases for target validation. We combine the experience and judgment of
our internal scientific leadership team with that of outside key scientific thought leaders to finalize the selection of a new therapeutic target to pursue.
believe that the proprietary software component of our Contour platform represents a compelling de novo design approach to drug
discovery because it increases the probability of successfully generating novel molecules that best fit into a protein binding site, as compared to the traditional human-guided computational methods.
Our Contour platform allows for the creation of novel, drug-like molecules by assembling synthetically viable fragments in a protein binding site using a high-resolution x-ray crystal structure or
homology model of the protein. As shown in Figure 26 below, our Contour platform does this by virtually growing drug-like molecules by assembling fragments in well-defined binding pockets.
Dynamic Fragment Selection, or DFS, is a novel component of Contour that uses the physical characteristics of the binding site in selecting complementary fragments during the growth process. Contour's
proprietary artificial intelligence experientially optimizes performance in subsequent iterations. We believe that DFS avoids the limitations of the standard growth and scoring approach since it
generates molecules by dynamically selecting only a subset of the fragment library that best matches the shape and features of a given pocket in the target binding site. Contour also provides a
3-dimensional representation of the small molecule bound to the binding site of the target protein and the optimal rotational position, or rotamer, of a fragment attached by a single, rotatable bond
to the rest of the compound. In greater than 90% of the cases, this representation is nearly identical to the structural information obtained by x-ray crystallography.
Figure 26: Our discovery process and illustration of the molecule assembly process by the Contour growth algorithm.
grown molecules are then scored by our proprietary empirical scoring function. The Contour scoring function is based on a directional contact model in which both distance and
orientation are used to characterize interactions. The directional model captures the close, basic molecular interactions such as