Virtual screening: search before you synthesize
Virtual screening uses computation to rank a large library so you only make and test the most promising fraction. When you have a structure of the target, the workhorse is molecular docking: the computer poses each candidate inside the pocket and a scoring function estimates how well it fits. When you lack a structure but have known actives, you can instead match candidates to a pharmacophore model — the abstract pattern of features a binder must present.
The appeal is reach: modern make-on-demand catalogs hold billions of synthesizable molecules, and you can dock through them without owning a single sample. The catch is honesty. A scoring function is a fast approximation of binding, not a measurement; it ranks well on average but is wrong on plenty of individual molecules. Virtual screening enriches your shortlist — it does not certify hits. Every computational pick still has to be made and tested for real.
DNA-encoded libraries: billions in one tube
A DNA-encoded library (DEL) takes combinatorial chemistry to an extreme. Each molecule is built on a strand of DNA that records its synthetic history like a barcode: every chemical step appends a unique DNA tag. Because the tag tells you the structure, you can pool millions or billions of distinct compounds into one tube and screen them all at once against immobilized protein.
- Synthesize the library by split-and-pool: at each step, split the pool, react each part with a different building block, attach a matching DNA tag, then recombine.
- Incubate the whole pool with immobilized target; wash away everything that does not stick.
- Elute the binders and amplify their DNA tags by PCR; sequence to read which structures were enriched.
- Re-synthesize the most enriched hits WITHOUT DNA and confirm activity in a normal assay.
DEL covers more compounds, more cheaply, than any physical screening library could. But the chemistry must be DNA-compatible, which limits the reactions available, and the DNA tag itself can influence binding. The final step is non-negotiable: a DEL output is only a hypothesis until the off-DNA molecule is made and shown to be active on its own.
Choosing — or combining — the methods
These approaches are complementary, not rival. Virtual screening is cheapest and works from a structure or known actives, but inherits all the blind spots of its scoring. DEL reaches the broadest swath of synthesizable chemistry but constrains the reactions and needs a protein you can immobilize. Classic HTS tests physical compounds under conditions closest to a real assay but costs the most per compound. Many groups now layer them: dock to focus a DEL, or HTS-confirm a virtual hit. The deciding factors are the same every time — do you have structure, can you immobilize the protein, and what can you afford to make.