Most hit-to-lead programs stall not because the hits were weak but because the structural basis for potency was never established. Designing the campaign around structural constraints from week one changes what questions you ask — and which hits you keep.
The Structural Deficit in Standard Hit-to-Lead Workflows
A common hit-to-lead entry point looks like this: a biochemical assay produces 40-80 confirmed actives from a diversity screen. Chemists rank them by IC50, flag obvious PAINS scaffolds, and select the 8-10 most potent, most drug-like compounds for initial analogue synthesis. The question asked in round one of chemistry is "what keeps or improves potency?" — and it is answered empirically, by synthesizing close analogues and measuring IC50 values.
The structural basis for why the original hit had any potency at all is often not established until round three or four — if it ever is. By then, the team has committed substantial synthesis resources to a chemical series based on empirical SAR patterns that may or may not translate once the actual binding mode is known. It is not unusual to crystallize a series lead at round four and discover that the potency driver is a water-mediated contact at a residue the team had not considered — and that the first 20 analogues synthesized were all trying to exploit a pharmacophore that was not the primary interaction.
SBDD hit-to-lead design is not simply "dock your hits and look at the poses." It is a campaign design philosophy that begins with establishing what the structure constrains before asking what the chemistry can change.
Structural Constraint Mapping Before Chemistry Begins
The first deliverable in an SBDD hit-to-lead campaign is a binding site constraint map: a characterization of which contacts are required, which are optional, and which are directional vectors that tolerate elaboration. This is derived from three sources simultaneously: the binding pose of the confirmed hit, comparison with co-crystal structures of related compounds at the same target (from public PDB data or internal series history), and first-principles analysis of the pocket geometry.
For a kinase hit-to-lead campaign, the constraint map typically looks like this. The hinge H-bond donor/acceptor pair is a required contact — scaffolds that lose it lose potency by at least one order of magnitude, regardless of other changes. The hydrophobic back pocket tolerates a range of lipophilic substituents, but the optimal vector angle from the scaffold is defined by the position of Thr/Leu gatekeeper residue at a specific backbone dihedral. The solvent-exposed region adjacent to the glycine-rich loop is a vector for selectivity-determining modifications, not potency modifications — changes there do not typically move IC50 unless they perturb solubility or ADMET behavior.
With this map in hand, the synthesis plan is constrained before the first compound is ordered. Modifications to required contacts are de-prioritized (unless there is a specific reason to challenge the hypothesis). Modifications to elaboration vectors are prioritized. Modifications to selectivity-modulating positions are scheduled for round two, after potency is established.
The Crystallography Checkpoint
The most important question in SBDD hit-to-lead campaign design is: at which point do you commit to a binding mode assumption without experimental confirmation? The answer varies by target confidence, but a pragmatic answer for most programs is this: docking-based binding mode assignment is sufficient for guiding the first round of chemistry — selecting which vectors to explore — but a crystallographic checkpoint is required before the third round. By the third round of chemistry, if the binding mode has not been confirmed, the SAR starts to drift.
Drift is the compound accumulation of chemistry decisions made on an unconfirmed structural hypothesis. Each individual decision is reasonable given available information; the problem is that they compound. By round three or four, the team has synthesized 30-40 compounds in a series direction that was plausible but not confirmed, and now reversing course costs more synthesis time than establishing the binding mode at round one would have.
The practical implication is that the hit-to-lead campaign design should budget for crystallography at two points: once for the confirmed hit itself (soaking or co-crystallization of the primary scaffold), and once for the round-two lead compound before committing to the third round of SAR. Programs that budget for chemistry but not for structural confirmation at the hit stage are optimizing for short-term synthesis velocity at the cost of structural understanding — and structural understanding is the asset that makes SAR efficiency possible downstream.
Selectivity as a Structural Design Problem, Not a Counter-Screen Problem
In a standard hit-to-lead workflow, selectivity profiling happens after potency is established: the top leads from round two or three are run against a selectivity panel (kinome panel, GPCR panel, protease panel), flags are noted, and the chemist tries to address them. This approach treats selectivity as a property to rescue, not a property to design.
SBDD hit-to-lead design inverts the sequence. The binding site constraint map includes a selectivity analysis: which residues in the target pocket are not conserved in the closest off-targets? What pocket geometry differences exist between the target and the most concerning selectivity liabilities? The answers to these questions define which chemical changes will favor selectivity versus penalize it — and those changes can be designed in from the start of SAR rather than applied as corrections at the end.
We're not claiming structural selectivity analysis replaces experimental profiling — it does not, and wet lab selectivity data remains the truth condition for any claim about a compound's selectivity window. What structural analysis does is reduce the number of selectivity-profile surprises at rounds three, four, and five, because the residue-level differences between target and off-target have been explicitly considered when planning which vectors to develop.
Scaffold Qualification and Exit Criteria
A hit-to-lead campaign needs exit criteria that are defined before chemistry begins, not discovered after. The criteria should be structural as well as potency-based. A scaffold that cannot achieve a confirmed co-crystal binding mode by round three, regardless of potency, is a scaffold that does not belong in a structure-guided program. This is a discipline that requires explicit commitment at campaign design: it feels expensive to exit a scaffold with a 100 nM IC50 because the binding mode cannot be confirmed, but the alternative — building a lead series on an unconfirmed structural hypothesis — is consistently more expensive downstream.
Consider a program targeting a novel PPI interface (a prototypical challenging target) where a fragment-based hit had confirmed binding by SPR (Kd ~150 μM) and was docked at the hot-spot sub-pocket with a plausible pose. Initial growth increased potency to 8 μM over three rounds. But the hot-spot residue assignment — which hot-spot is the primary anchor? — could not be resolved by docking alone given three structurally similar hot-spot cavities within 8 Å of each other. Rather than continue chemistry, the right call was to attempt soaking with a panel of elaborated fragments to establish which sub-pocket was actually occupied. That experiment takes two weeks and delays chemistry by two rounds. It saves the program from six months of chemistry in the wrong sub-pocket.
Defined scaffold qualification criteria also help with a less obvious problem: the pull of the "best current compound." The best current compound, measured by IC50, tends to absorb disproportionate synthesis effort even when its structural properties are not optimal. Explicit criteria around binding mode confidence, LE trajectory, and selectivity profile prevent the campaign from narrowing prematurely to a single scaffold before its structural liabilities are understood.
Integrating ADMET from Round One
ADMET considerations in a traditional hit-to-lead workflow enter at the lead selection stage — after the potency and selectivity work is done. In an SBDD-designed campaign, they enter at constraint mapping. Not because early-stage ADMET predictions are highly accurate — they are not — but because certain structural features that disfavor ADMET are also visible in the binding pose. A hit that achieves potency through a substituent that adds a metabolic soft spot in a CYP3A4-labile position can sometimes be redesigned using a different elaboration vector, if the structural analysis has been done first. Identifying that option in round one is more efficient than discovering the CYP3A4 liability in round four and restarting.
SwissADME, pkCSM, and similar rapid ADMET prediction tools are adequate for early triage decisions at the fragment and hit stage — not for generating precise numbers, but for flagging compound classes that routinely fail at specific ADMET parameters. Incorporating this alongside the binding site constraint map at campaign design creates an integrated prioritization matrix, not a sequential filter stack.
For the detailed methodology behind how these structural constraint maps are built for different target classes, see the Capabilities page. For a discussion of how fragment starting points interface with hit-to-lead campaign design, see Why Fragment Libraries Outperform HTS Sets.