Structural biology resolves the 3D atomic shape of a drug target and how a molecule binds it, using X-ray crystallography, cryo-EM, NMR, and biophysics. You need it in discovery, mainly during hit-to-lead and lead optimization, to guide structure-based design. On BioBridgeX you source and compare qualified structural biology CROs under one contract.
Structural Biology CROs (8)
OpenEye, Cadence Molecular Sciences
Unclaimed · public records
CRO · Hit-to-Lead, Lead Optimization, Computational / AI-Driven Discovery
What is structural biology and when do you need it in drug development?
Structural biology is the work of seeing your target at atomic resolution: the three-dimensional shape of a protein, the pocket a small molecule has to fit, and the exact contacts a ligand makes when it binds. The classic methods are X-ray crystallography (still the workhorse for small-molecule co-crystal structures), cryo-electron microscopy (now the tool of choice for large complexes, membrane proteins, and targets that will not crystallize), solution NMR for smaller proteins and fragment work, and a layer of biophysics (SPR, ITC, thermal shift, MST, HDX-MS) that measures how tightly and how fast things bind. Together they answer a question medicinal chemistry cannot answer on its own: not just whether a compound is active, but why, and where on the protein it sits.
In a typical small-molecule program, structural biology earns its keep in two places. The first is at hit-to-lead, when you want to confirm a screening hit actually binds the intended pocket and not some artifact site, and to see its binding mode before you commit chemistry to it. The second, and the heavier spend, is across lead optimization, where co-crystal or cryo-EM structures of your series guide each design cycle: which substituent to grow into an unfilled pocket, where a clash is costing you potency, how to pick up selectivity against a related target. This is the engine room of structure-based drug design, and a program running structure-blind is usually slower and burns more analogs to get to the same place.
Not every program needs it, and that is worth saying plainly. Phenotypic campaigns, some antibody-discovery work, and targets with no tractable structure can advance for a long time without a single solved structure. But for kinases, proteases, GPCRs, nuclear receptors, protein-protein interaction targets, and most fragment-based or PROTAC efforts, structural data is close to mandatory. The buyer's real decision is not whether structural biology is useful, it is which method fits this target, whether to run it in-house or outsource it, and how fast a CRO can turn a soakable crystal system or a cryo-EM map into design-ready structures.
What does a structural biology CRO actually do?
The scope is broader than just collecting data on a synchrotron. A capable structural biology CRO usually owns the chain from construct to structure: designing and screening protein constructs, expressing and purifying enough well-behaved, monodisperse protein to work with, then crystallization (sparse-matrix and targeted screens, optimization, and crucially a soaking or co-crystallization system that lets you turn compounds around quickly), data collection at a synchrotron, and structure solution and refinement to a deposited-quality model. For cryo-EM the path runs through sample vitrification, screening, high-resolution data collection, and the heavy image-processing and model-building that follows.
Two parts of that chain are where programs actually stall, and they are worth probing hard before you sign. One is getting tractable protein: a target that expresses poorly, aggregates, or refuses to behave can sink a timeline no matter how good the crystallography group is, so construct engineering and protein production capability matter as much as the diffraction work. The other is throughput on ligand structures: in lead optimization you are not asking for one heroic structure, you are asking for many co-structures fast, so a dependable soaking system and a turnaround measured in days or a couple of weeks per batch is what keeps a design cycle moving. Alongside the core methods, most groups also run the biophysics that supports them, SPR and ITC for affinity and kinetics, thermal shift and MST for binding confirmation, and increasingly support computational and AI structure prediction (AlphaFold and related models) to bootstrap a starting model when no experimental structure exists yet.
How do you choose a structural biology CRO?
The decision that matters most is method-and-target fit, not the size of the logo. A group that produces beautiful soluble-protein crystal structures may be the wrong call for a membrane GPCR or a large complex that needs cryo-EM, and a fragment campaign needs a high-throughput soaking system that not every shop runs well. Ask for relevant, recent structures in your target class, and find out whether the scientists who would run your project have actually solved this kind of target before, ideally with deposited PDB or EMDB entries you can look at. Then weigh these against your timeline and your contract terms:
Quality and GxP status: structural biology in discovery is research-grade, not GLP, so the right quality signal is scientific rigor and reproducibility, not a GLP certificate. Look for documented protein characterization (purity, monodispersity, identity), sound data-collection and refinement statistics (resolution, Rwork/Rfree, map quality, validation reports), and a track record of deposited structures.
Capacity and lead time: confirm real availability, not a sales promise. Ask about current queue, expected turnaround for protein production versus a batch of co-structures, and whether they can sustain the throughput a lead-optimization series needs (many ligand structures over months, not one a quarter).
Modality and indication fit: match the CRO to your target type (soluble enzyme, kinase, membrane protein, GPCR, protein-protein interaction, nucleic-acid complex) and to your modality, whether that is small molecule, fragment, PROTAC degrader, peptide, or a biologic where epitope mapping is the question.
Region and regulatory track record: structural data rarely goes into a regulatory filing directly, so the live questions are time-zone and communication fit, synchrotron and cryo-EM facility access in their region, and whether their data and reporting standards hold up when your downstream teams rely on them.
Data quality and deliverables: pin down exactly what you receive. Refined coordinates and structure factors (or cryo-EM maps), validation reports, the construct and protein details, raw versus processed data, and clear electronic records. Vague deliverables are where structural projects disappoint.
IP and confidentiality: settle who owns the structures, the constructs, and any platform-derived methods before work starts, and confirm a strong CDA, because the target itself and your compound binding modes are often the most sensitive information in the program.
Frequently asked questions
What is the difference between X-ray crystallography and cryo-EM, and which do I need?
X-ray crystallography needs your protein to form an ordered crystal, and it remains the fastest, highest-throughput route for small-molecule co-structures, which is why it dominates lead optimization for soluble, crystallizable targets like many kinases and proteases. Cryo-EM does not need crystals and shines on large complexes, membrane proteins, and targets that simply will not crystallize, and its resolution has improved enough to support drug design on many of those. The practical rule: if your target crystallizes and soaks well, crystallography is usually cheaper and faster per ligand; if it does not, or it is a big membrane complex, cryo-EM is the path. A good CRO will tell you honestly which fits your target rather than selling the method they happen to own.
At what stage of discovery do I need structural biology?
Most programs first call on it at hit-to-lead, to confirm a screening hit binds the intended pocket and to see its binding mode, and then lean on it heavily through lead optimization, where co-structures of your chemical series guide each design cycle. Fragment-based programs need it even earlier, since fragment hits are weak and structural data is often the only way to see where and how they bind. Phenotypic campaigns and some antibody work can run a long way without a solved structure. The honest test is whether your chemistry decisions would be better with a picture of the binding site; for most defined-target small-molecule and PROTAC programs, the answer is yes.
How long does it take to get a co-crystal structure?
It depends almost entirely on whether a tractable protein and a working crystal system already exist. If a CRO already has a validated, soakable crystal form for your target, a batch of ligand co-structures can come back in roughly one to a few weeks. If you are starting cold, the front end (construct screening, protein production, finding and optimizing crystallization conditions, establishing a soaking or co-crystallization system) is the real timeline and commonly runs months, with difficult or membrane targets longer or sometimes never crystallizing at all. Ask any vendor to separate one-time system-setup time from steady-state per-structure turnaround, because those are very different numbers and the second is what governs your design cycles.
Does structural biology work need to be done under GLP?
Generally no. Structural biology in discovery is research-grade work conducted under good scientific practice, not under GLP, GMP, or GCP, because its output guides design decisions rather than supporting a safety filing. What you should expect instead is scientific rigor: well-characterized protein (purity, identity, monodispersity), sound data-collection and refinement statistics, standard validation reports, and clean, traceable electronic records. If a vendor offers GLP for structural work, treat it as a sign to ask what specifically is being claimed, since paying GLP premiums for exploratory discovery structures is usually wasted money.
Who owns the structures and constructs when I outsource structural biology?
Settle this in writing before any work starts, because the structures of your target bound to your compounds are among the most sensitive assets in the program. In a well-structured agreement the buyer owns the solved structures, the data, and typically the constructs generated for the funded work. The items to watch are vendors with proprietary expression or crystallization platforms who may claim rights to platform-derived methods, and the scope of any future publication or deposition. Confirm a strong confidentiality agreement covering the target identity itself, and make data and coordinate transfer terms explicit so you are not negotiating for your own structures later.
Can AlphaFold or AI structure prediction replace an experimental structural biology CRO?
Not for drug design, not yet. AlphaFold and related models are genuinely useful for generating a starting model, prioritizing constructs, and reasoning about a target before you have experimental data, and they have changed how programs begin. But predicted models do not reliably capture ligand binding modes, induced-fit changes, water networks, or the small shifts in a pocket that determine potency and selectivity, which is exactly the information lead optimization runs on. The strongest programs use AI prediction to accelerate the experimental work, not to replace the co-crystal or cryo-EM structures that show how your actual compound binds your actual target.
Source Structural Biology with one contract
Compare transparent quotes from qualified Structural Biology CROs, then contract once. Free for buyers.
