Drug Design Process

Our drug design process begins by computationally enumerating and analyzing billions of synthetically tractable molecular design ideas based on physics-driven structural hypotheses.

We triage the vast chemical space of virtual molecules by using property prediction models and structural filters to identify molecules with desirable drug-like properties. We then subject these molecules to accurate in silico assays for binding, selectivity and ADME properties to identify the best candidates for synthesis and in vitro assay. Our iterative process includes the following steps:”

Design Paradigm

Hypothesis & Ideation

Simulations identify binding hotspots, induced pockets, and optimizable regions. We track all hypotheses to measure progress toward TPP and to make continual improvements.

Enumeration

Reaction-based enumeration, molecular breeding, and generative modeling explore chemical space based on a specified hypothesis. Our curated chemical building block database ensures that the chemical constituents can be purchased to synthesize the enumerated molecules.

Fingerprint Filters

Enumerated chemical space is reduced based on drug-likeness and chemical similarity. These fast calculations eliminate the most unsuitable molecules.

Property Filters

In-house QSAR models using proprietary curated datasets predict ADME properties such as solubility, permeability, hERG, and clearance. Only the most suitable molecules progress past this stage.

Structure-Based Filters

Docking, shape overlap, and contact maps account for general fit to the protein pocket and desirable interactions based on design hypotheses.

Quick Simulations

Short molecular dynamics simulations and MM-GBSA account for light protein flexibility, protein-ligand interactions, and desolvation. This step eliminates enough molecules for us to move into rigorous free energy simulations.

Rigorous Simulations

Rigorous binding free energy simulations account fully for the thermodynamic contributions to binding (enthalpy and entropy). Additionally, advanced molecular dynamics techniques are applied when long-timescale motions are involved, such as allosteric modulation and signaling.

Synthesis and Assays

We prioritize molecules based on a multi-parameter analysis guided by the TPP. We also account for synthetic accessibility, time to results, and costs to progress our synthesized compounds through the assay cascade.

Iterate

We collect, analyze and discuss what we learn from data to determine the focus of the next iteration. We track all data our Project Dashboard, with real-time monitoring and reporting. Throughout Psivant we make visible our progression toward TPP to ensure transparent data-driven decisions.