Screening Capabilities by Target Class

The ATP-binding site is highly conserved across the kinome — selectivity comes from hinge geometry, the DFG loop conformation (DFG-in vs. DFG-out), gatekeeper residue identity, and allosteric pocket availability. Fragment campaigns biased toward hinge-binders with defined vectors into the hydrophobic back pocket produce structurally differentiated hits. Moleculepath routinely profiles top-ranked hits against a computational kinase selectivity model before shortlist delivery — compounds with predicted pan-kinase activity are flagged before synthesis resources are committed.

For targets with published allosteric sites (type II kinase inhibitors, myristoyl pocket), separate fragment campaigns against the allosteric site can be run in parallel. DFG loop conformation treatment requires careful structure preparation — crystallographic DFG-out conformations should not be docked against DFG-in fragment libraries, and vice versa.

Available GPCR crystal structures now cover multiple families, including Class A, Class B, and Class C representatives in both active and inactive states. Orthosteric site docking is feasible when co-crystal structures are available in the appropriate activation state for the desired pharmacology. Allosteric and bitopic fragment strategies require conformational sampling — the extracellular loops and transmembrane helical bundle flexibility must be accounted for in the docking protocol.

Program design for GPCR targets begins with an explicit discussion of structure quality, activation state, and the pharmacological objective. Agonist vs. antagonist vs. inverse agonist campaigns each require different structural inputs. Homology models from close Class A orthologs are acceptable starting structures when key binding site residues are conserved; confidence metrics are characterized and included in the campaign design proposal.

Serine, cysteine, aspartyl, and metalloprotease subfamilies each have well-defined S1 and S2 pocket architectures that dictate pharmacophore requirements. The catalytic mechanism affects covalent vs. non-covalent fragment strategy — cysteine proteases are amenable to electrophilic fragment campaigns; serine proteases require non-covalent hit identification followed by separate warhead conjugation studies.

Selectivity in the protease class is typically driven by S1 pocket geometry and the identity of the gating residue at the S1/S2 boundary. Fragment libraries for protease campaigns are pre-filtered for S1 compatibility and non-covalent binding mode — PAINS filters are applied more stringently here due to the prevalence of false positives from reactive fragments against catalytic residues.

Bromodomain, chromodomain, and PWWP-domain readers recognize post-translational histone modifications through shallow, relatively flat binding cavities. The BET bromodomain family is the most structurally characterized; BRD4-BD1 and BD2 have distinct pocket geometries that support selectivity between paralogs. Non-BET bromodomains often have narrower acetyl-lysine binding channels that require smaller, more precisely shaped fragments.

Epigenetic reader programs typically benefit from competitive displacement analysis against known histone peptide mimetics to confirm binding mode before advancing hits. Selectivity profiling against the broader bromodomain family is available computationally using structural comparison of the acetyl-lysine channel geometry across paralogs.

PPI interfaces are typically large and flat — the buried surface area (800–2,000 Ų) is not ideal for single-fragment occupancy. Successful PPI inhibitors typically target hot-spot sub-pockets identified by alanine scanning data or cryo-EM complex structures. Fragment campaigns against PPIs require explicit hot-spot mapping before library selection — running a standard docking screen against the flat interface produces low-quality results regardless of protocol quality.

Moleculepath's PPI workflow begins with hot-spot analysis: structural comparison of the interface geometry, identification of surface concavities compatible with fragment binding, and literature review of known small-molecule mimetics for the interaction class. Fragment campaigns are then designed around the identified hot-spot sub-pockets, not the full interface.

Covalent drugs are no longer a last resort. Structure-guided warhead selection — matching electrophile reactivity to the target cysteine microenvironment — is now a tractable hit identification strategy for the right target class. Covalent campaigns require the identification of an accessible cysteine (or lysine, tyrosine, or histidine for other warhead classes) within or adjacent to a druggable sub-pocket.

The Moleculepath covalent workflow: (1) cysteine accessibility and microenvironment analysis from the target structure; (2) non-covalent fragment docking to identify scaffolds that position a reactive vector toward the target cysteine; (3) warhead class selection based on the cysteine microenvironment (acrylamide, vinyl sulfone, cyanoacrylamide, or chloroacetamide) matched to the desired reactivity profile; (4) docking of warhead-conjugated fragments for pose confirmation. Chemically reactive screening sets are not used — reactivity is designed in after structural analysis, not selected from a library of pre-made electrophiles.