Twist launches cell-free screening for AI-designed antibodies

By Rebecca Stone

Twist Bioscience launched Express Screening Antibodies on October 28 with starting pricing of $87.75 per variant and a five-business-day cell-free turnaround, creating a synthesis-to-binding decision gate for antibody candidates designed with artificial intelligence (AI).

What problem does the service solve?

Twist, which operates in South San Francisco and Wilsonville, Oregon, says the service addresses a gap between computational candidate generation and experimental testing. Drug developers once examined about 100 antibodies at a time; AI models running on pharmaceutical supercomputers now propose thousands, according to CEO and co-founder Emily M. Leproust, Ph.D.

Leproust said the service brings “speed and scalability to early-stage characterization.” Its intended workflow is to test broad sequence sets, retain experimental data and direct mammalian-expression budgets toward candidates that show the strongest binding.

At about 100 antibodies, buyers can choose among several vendors, Leproust said. At 1,000 or 10,000, she argued that few suppliers can synthesize, produce and test that many candidates. Her blunt assessment—“forget it. We’re the only game”—states a competitive claim, not a benchmark; customers still need evidence on quality, reproducibility and capacity.

In a July factory tour, Leproust said Twist runs DNA synthesis, production and testing in-house. She used a chess analogy in which Twist supplies the board and gives developers more pieces, hence more opportunities to find a viable biologic. The commercial logic depends on quality remaining consistent across campaigns.

How does the cell-free workflow operate?

Customers submit variable-region sequences, the antibody portions that recognize a target. Twist synthesizes the heavy- and light-chain genes as linear DNA on its silicon platform, removing a cloning step before expression.

Expression uses ALiCE (Almost Living Cell-free Expression), a protein-production system from LenioBio in Düsseldorf, Germany. LenioBio makes proteins for customers and uses ALiCE in its own laboratories. The system makes proteins without intact cells in a lysate; a 2025 study with LenioBio co-authors describes that lysate as an extract of broken-open tobacco BY-2 plant cells.

• Twist expresses the DNA in 50- or 100-microliter reactions on 96-well plates, then purifies assembled antibodies from the lysate in one Protein A affinity-capture step.
• Output options include full-length human immunoglobulin G1 (IgG1) or two smaller fusion formats: scFv-Fc, a single-chain variable fragment joined to an antibody’s Fc (fragment crystallizable) tail, and VHH-Fc, a single-domain binder derived from camelid heavy-chain antibodies joined to that same tail.
• The launch materials list output quantities of 5 to 20 micrograms.
• Product-sheet averages are about 10 to 15 micrograms of IgG from a 50-microliter reaction and 15 to 20 micrograms from 100 microliters. Fusion formats yield about 10 to 30 micrograms.

Researchers can route material directly into surface plasmon resonance (SPR), which measures how quickly and tightly each construct binds its target. An SPR run uses 5 to 15 micrograms per construct, while Twist reports yield for each sequence as an early screening signal.

How quickly does the service deliver?

Cell-free antibodies ship five business days after Twist receives the order. Adding SPR binding data extends delivery to eight to ten business days. Twist specifies that each figure reflects its fastest configuration; format, order size and sequence complexity can lengthen it.

Mammalian production takes longer before any assay begins: 8 to 11 business days in HEK293, a human cell line derived from embryonic kidney tissue in the 1970s and widely used to make research proteins, and 11 to 14 days in CHO (Chinese hamster ovary), the line drugmakers most often use to manufacture antibody drugs.

The faster route produces less material at lower purity. Twist lists cell-free purity at about 70%, compared with about 95% for most HEK293 and CHO formats. Published pricing starts at $87.75 per variant.

The launch announcement describes assessment of thousands of antibodies per day, while the product sheet places campaign scale at hundreds to thousands of candidates triaged. The reported material gives no campaign-level failure rates or independent throughput measurements.

How convincing is the validation?

Twist ran a head-to-head SPR comparison of eight antibodies against seven antigens, expressing each as cell-free IgG and as a reference antibody made in HEK293 cells. It compared equilibrium dissociation constants (KD), a measure of binding strength for which lower values indicate tighter binding.

The two sets of KD values correlated at a Pearson r of 0.973 on a log-log scale. Seven of the eight cell-free values fell within threefold of their HEK293 counterparts, and the tightest binder measured 12.6 picomolar.

That result supports rank consistency in a small, vendor-run test, but it does not establish equivalence across formats or larger sequence sets. The reported validation identifies Twist as the provider, gives no independent-replication detail and names no funding source. It also provides no head-to-head results for the two fusion formats.

Where does cell-free fit in the portfolio?

Selected candidates can advance into HEK293 or CHO expression. A 1-milliliter HEK293 run yields an average of 180 micrograms, providing material for developability assays that test thermal stability, aggregation, self-interaction and nonspecific binding. Together, these assays indicate whether a binder can withstand processing and avoid problematic molecular interactions.

Twist offers those developability assays only for its two mammalian systems. SPR affinity measurement is the only one of its six binding assays that lists cell-free material, so the cell-free stage functions as a front gate rather than a substitute for full characterization.

For budget owners, the starting price of $87.75 creates a low-cost first decision point before mammalian production. It does not remove the need to compare binding purity, expression yield, data quality and total campaign cost when deciding which candidates merit scale-up.

How does Twist connect to AI workflows?

Amazon Bio Discovery, introduced by Amazon Web Services (AWS) in April, lists Twist, Ginkgo Bioworks and A-Alpha Bio as partners that synthesize and test its designs.

In May, Benchling added in-notebook ordering for gene synthesis, antibody expression, and binding and developability assays from Twist, Adaptyv and Ginkgo. On September 16, Twist joined Eli Lilly’s TuneLab AI drug-discovery platform as a provider of antibody characterization data, including data for AbLab, a model that predicts antibody developability.

These integrations give Twist three routes into AI-led discovery. Their practical value depends on producing comparable data across expression systems and supporting portfolio decisions at campaign scale, rather than simply moving candidate lists between vendors.

The next procurement test is whether Twist can reproduce its 0.973 binding correlation in customers’ larger, sequence-diverse campaigns and convert those screening data into fewer, better-developed antibody leads.