Nucleic Acid Synthesis A Guide for 2026

Anyone holding a sequence file has a practical question to answer. The question is whether the construct can be made quickly and cleanly, in a form that does not break the rest of the workflow.

For most research teams, nucleic acid synthesis has become the manufacturing layer beneath much of their work. It supplies primers and genes, and it also underlies work that is treated as routine until a construct fails synthesis, assembly, or QC. That work includes the following.

• CRISPR editing
• mRNA programs
• antisense work
• screening libraries
• metabolic engineering
• day-to-day molecular biology

The hard part in 2026 is choosing the right synthesis path, anticipating where the sequence will misbehave, and reducing failure before any wet-lab work starts. Access is no longer the difficulty. Teams that treat synthesis as a simple procurement step usually pay for it later in redesign cycles, assembly delays, and QC surprises.

The Engine of Modern Biotech Nucleic Acid Synthesis

A custom nucleic acid sequence now sits at the front of almost every serious biotech program. One group needs a guide RNA set for a CRISPR screen, and another needs a codon-adjusted gene for expression testing. A therapy team needs DNA templates that will support an efficient RNA workflow. The endpoints differ but the dependency is the same. Someone has to turn a designed sequence into real material that behaves in the lab.

The importance of that capability is clear from how the field evolved. The nucleic acid therapeutics field emerged from pioneering work in the late 1970s, and in 1978 synthetic oligonucleotides were used to inhibit viral replication, which established proof-of-concept for nucleic acids as drugs. The combination of therapeutic interest with automated synthesis chemistry in the 1980s made it economically feasible to explore these applications at scale. This is one reason nucleic acid therapeutics developed into a major therapeutic class, as described in the NCBI overview of nucleic acid therapeutics.

What teams have to optimize

In practice, most synthesis decisions come down to a set of trade-offs:

• Sequence length: Short oligos and longer constructs behave very differently in synthesis and assembly.
• Error tolerance: A screening library can tolerate a different QC model than a final therapeutic template.
• Turnaround pressure: Some programs need a few validated constructs. Others need many iterations.
• Downstream use: PCR, cloning, IVT, genome editing, and therapeutic development all place different demands on the starting material. IVT is in vitro transcription, which makes RNA from a DNA template.

Those trade-offs matter because synthesis has become a chain of decisions. The chemistry or enzymology is only one layer. Assembly strategy, purification burden, repeat content, secondary structure, and quality control all shape whether a sequence is useful after delivery.

Practical rule: The best synthesis workflow is the one that minimizes redesign. The workflow that looks fastest on the ordering page is not necessarily the best one.

Why computation now belongs at the start

The strongest teams pre-screen the design instead of waiting for the vendor to say that a sequence is hard to make. They remove unstable repeats when possible and rethink problematic regions. They also choose workflows that fit the molecule, and they avoid forcing the molecule into a default ordering pipeline.

That is the central shift. Wet-lab synthesis still does the physical work, but digital design increasingly determines whether the project moves smoothly or stalls. Once nucleic acid synthesis is treated as an engineering problem, the choices become clearer.

Chemical vs Enzymatic A Tale of Two Synthesis Strategies

The practical choice is whether the synthesis method matches the sequence, the assembly plan, and the failure modes the team can afford. Computational pre-screening matters here because many of the sequences that cause trouble in synthesis also cause trouble in assembly, amplification, and QC later.

Why phosphoramidite chemistry still matters

Phosphoramidite chemistry remains the default for good reasons. It is mature and widely available. It is also tightly integrated into vendor pipelines for primers, probes, antisense oligos, guide components, and other short DNA or RNA products.

Its default status also has a historical basis. The chemical synthesis of nucleic acids began in the 1950s. The field changed when Marvin Caruthers developed phosphoramidite methods between 1977 and 1982, which enabled automated solid-phase synthesis at useful scale, as described in the University of Edinburgh Engineering Life historical account.

For short formats, chemistry is still hard to beat. Turnaround is familiar, modification options are broad, and many labs already have downstream workflows built around standard oligos.

Where chemical synthesis becomes expensive in practice

The weaknesses of chemical synthesis appear as sequence length and complexity increase. A 20mer primer and a difficult several-hundred-base fragment behave as different product classes.

A 2025 PMC review on enzymatic DNA synthesis notes that phosphoramidite workflows are generally constrained beyond about 200 bp, with cumulative errors rising and repetitive motifs performing poorly. The same review also describes the solvent burden and purification overhead that make longer chemical products more expensive and less convenient to handle at scale.

That cost shows up in redesign cycles, failed assemblies, and extra screening as well as on the invoice. If a sequence has repeats, strong secondary structure, or awkward GC structure, a computational check before ordering can save more time than any downstream cleanup step.

What enzymatic synthesis changes

Enzymatic DNA synthesis (EDS) uses enzymes to add nucleotides in place of repeated chemical coupling. In current implementations, that often means engineered terminal deoxynucleotidyl transferase (TdT) systems, which extend a DNA strand one nucleotide at a time under control.

The appeal is straightforward. As summarized in that same PMC review, enzymatic platforms have reported high per-cycle fidelity, lower truncation burden, solvent-free production, and prototype-scale generation of longer constructs on faster timelines than traditional chemistry. Those are meaningful advantages when the bottleneck is getting sequence-correct material that assembles cleanly.

In practice, EDS is a better fit for a growing subset of jobs. EDS is not a universal replacement. Longer constructs, difficult motifs, and workflows where assembly dominates the schedule are the clearest examples of that subset.

A side-by-side comparison

Metric	Phosphoramidite (Chemical)	Enzymatic DNA Synthesis (EDS)
Core mechanism	Solid-phase chemical coupling using phosphoramidites	Enzyme-mediated nucleotide addition using engineered TdT variants
Historical status	Long-established industry standard	Emerging commercial approach
Practical strength	Reliable for many short oligos and modified products	Better fit for some longer constructs and lower truncation workflows
Length behavior	Performance drops as fragments get longer, especially on difficult sequence content	Reported progress toward longer direct synthesis in the PMC review
Error profile	Cumulative errors and truncations become harder to manage with length	Reported high per-cycle fidelity in the same review
Repeats and difficult motifs	Often problematic	Potentially better suited, depending on platform and sequence
Solvent burden	Requires organic chemistry workflow and cleanup	Reported solvent-free synthesis approach
Downstream assembly	Often pushes projects into multi-step assembly	Can reduce assembly burden if longer high-quality material is produced directly

How to choose the method

Use chemical synthesis for short oligos, standard modifications, predictable vendor support, and compatibility with established lab workflows. That still covers a large share of day-to-day molecular biology, from primers to probes to many library-building inputs. For a quick baseline on short synthetic building blocks, this overview of what an oligo is and how it is used in the lab is a useful reference.

Use enzymatic synthesis when the design itself is the source of risk. Longer targets, repetitive regions, and projects where assembly and screening consume more time than ordering are the cases where EDS becomes attractive.

The better decision process starts before purchase. Simulate the sequence. Flag repeats, extreme GC windows, hairpin-prone regions, and assembly breakpoints. Then choose the synthesis route that reduces total experimental friction across the whole project. First-step turnaround is only one part of that. This is the shift shaping modern nucleic acid synthesis. Wet-lab chemistry still matters, but digital design now decides which chemistry is likely to work.

From Oligos to Genes Mapping Common Synthesis Workflows

The method matters, but researchers often approach nucleic acid synthesis through workflows. The request usually takes the form “I need this mutant, this guide set, this gene, or this RNA template.”

Oligo synthesis as the base layer

Most workflows still start with short oligonucleotides. The following all sit in this category.

• Primers
• probes
• barcodes
• adapters
• antisense sequences
• assembly fragments

This explanation of what an oligo is is a useful baseline for anyone new to the terminology because it connects the simple definition to real laboratory use.

Short oligos are the building blocks of molecular biology and are usually inputs to the final product. They make the final product possible.

This matters when planning procurement. A primer set for routine PCR can often be treated as disposable infrastructure. An oligo pool intended for library construction cannot. The second case needs tighter thinking around sequence balance, amplification bias, and QC before anything is ligated or transformed.

Gene synthesis in practice

A common gene synthesis workflow starts with designed oligos that tile across a target sequence. Those oligos are assembled, amplified, cleaned up, and then sequence-verified before cloning or direct functional testing.

The practical weak points are familiar:

• Assembly junctions: Overlaps that look fine in software can still produce uneven assembly behavior.
• Repetitive regions: Repeats complicate oligo design and often create downstream verification headaches.
• Sequence context: High or uneven GC regions can make amplification and cleanup less forgiving.

When this workflow works well, the synthesized gene arrives as a stable, verified input for expression or editing. When it goes poorly, the sequence may still exist on paper but not in a form that survives assembly, cloning, or QC.

In vitro transcription for RNA programs

RNA workflows usually begin with a DNA template. That template might come from a synthetic gene, a PCR product, or an assembled DNA fragment prepared for transcription.

For teams building RNA constructs, the upstream DNA choice has consequences. Template quality affects transcription behavior, impurity burden, and how much troubleshooting gets pushed downstream into RNA analytics. Many groups therefore find that DNA made successfully may still be unfit for IVT.

If the downstream product is RNA, treat the DNA template as a manufacturing intermediate.

PCR-based assembly and mutagenesis

PCR-based assembly remains one of the most flexible tools in the lab. It is fast to prototype with and adaptable for mutagenesis. It is also realistic for small-batch design cycles where ordering a fully assembled gene every time would slow iteration.

A standard mutagenesis path usually looks like this:

1. Define the target region and decide whether you’re introducing a point mutation, a short insertion, or a larger swap.
2. Design primers with enough local stability to support amplification while avoiding unnecessary secondary structure.
3. Run amplification and cleanup, then move immediately into verification.
4. Transform or assemble onward only after checking that the product length and basic quality look right.

Small details matter here. Consistency in thermal cycling and sample handling helps, and consumables contribute to it.

Library construction as a different class of problem

Library workflows look similar on the surface and behave differently in practice. The question changes from whether one construct assembled correctly to whether many intended variants survived synthesis, amplification, cloning, and selection without major distortion.

That changes the planning logic. The priority shifts from one perfect molecule to representational fidelity across the pool. Sequence diversity, amplification bias, and readout strategy matter as much as the synthesis order itself.

Confronting Errors The Reality of Synthesis Fidelity and QC

Every synthesis method produces errors. The question that matters is where the errors enter, how visible they are, and whether the QC plan matches the risk of the workflow.

Teams often underestimate this because ordering DNA feels transactional. A sequence is submitted, material is received, and work moves on. Synthesis is not deterministic in the way people often assume. Molecules truncate. Bases misincorporate. Assemblies create mixed populations. Amplification can distort what was already imperfect.

The error modes that show up most often

In day-to-day work, the recurring categories are easy to recognize:

• Truncations: Incomplete products are especially relevant when coupling efficiency or extension performance drifts.
• Substitutions: Single-base changes can remain undetected until a functional assay fails.
• Insertions and deletions: These are often more damaging than substitutions because they can alter reading frame or disrupt regulatory elements.
• Population heterogeneity: Pools and assembled products may contain the intended sequence plus closely related byproducts.

The weight of each category varies between programs. A disposable screening oligo can tolerate a different impurity profile than a sequence destined to anchor a therapeutic workflow.

QC has to match the construct type

Single constructs and pooled constructs need different verification logic.

For an individual plasmid insert, many labs still start with Sanger sequencing because it answers the immediate question quickly, which is whether the construct is correct. For pooled libraries, that approach does not scale. Understanding composition and dropout requires next-generation sequencing or another population-aware assay.

Purity is a measured property tied to the assay used, the impurity classes of interest, and the decision that follows. A certificate label alone does not convey it.

Don’t outsource all judgment to the vendor COA

Vendor documentation is useful, but it does not describe the whole sample in the experiment. Teams should read certificates of analysis critically, especially when moving from routine oligos to material that influences expensive downstream work. This guide to a certificate of analysis COA is a good reference for understanding what a COA can confirm and what it cannot.

QC mindset: Verify the feature that can really break your experiment, even when another feature is easier to measure.

That means checking the assembled junction if assembly is the risk. It means sequencing the edited region if mutagenesis is the risk. It means evaluating distribution across a pool if library representation is the risk. Generic pass-fail quality checks often miss the reason a project fails two weeks later.

What doesn’t work

Two habits create avoidable trouble.

First, teams sometimes skip post-synthesis verification because the sequence was “vendor verified.” This is usually fine until the construct enters a more sensitive assay where a minor defect becomes expensive. Second, teams often apply the same QC template to every synthesis job. A library, a therapeutic template, and a quick cloning fragment should each be treated differently.

The effective approach is to tie QC depth to downstream consequence. If failure is cheap, verify lightly. If failure propagates through many assays, verify aggressively and early.

Designing for Success With Computational and AI Tools

Most synthesis failures are easier to prevent than to diagnose. By the time a sequence has failed assembly, produced a poor library, or created inconsistent expression data, the cost of a design mistake that could have been caught upstream has already been paid.

That is why modern nucleic acid synthesis increasingly starts in software.

Design for manufacturability applies to DNA too

Engineers in other fields use design-for-manufacture principles because producing a thing is easier when the design respects the process. DNA and RNA follow the same logic. A sequence can be biologically correct and still be awkward to synthesize, amplify, assemble, or express.

Computational pre-screening helps teams catch problems such as:

• Problematic repeats: These can destabilize synthesis and complicate assembly.
• Secondary structure hotspots: Hairpins and self-complementary regions can interfere with amplification or processing.
• Extreme local composition: GC imbalance can create uneven behavior even when average content looks acceptable.
• Variant placement issues: A mutation that seems trivial may create a synthesis or assembly bottleneck because of surrounding context.

The practical value is straightforward. Sequence redesign moves to the cheapest stage, which is before ordering.

Optimization is more than codon choice

Codon optimization is useful but is only one layer. In many projects, codon changes interact with synthesis constraints, RNA structure, translation behavior and cloning strategy. Optimizing blindly for one objective can create trouble somewhere else.

A good computational workflow weighs multiple constraints at once:

1. Preserve biological intent, including coding sequence or regulatory function.
2. Remove synthesis liabilities such as repeats and difficult motifs when possible.
3. Support downstream assembly as well as raw synthesis.
4. Anticipate assay context, especially if the construct will feed IVT, editing, or pooled screening.

That matters even more when teams move beyond canonical DNA design. An underserved area in nucleic acid synthesis is computational modeling for non-standard backbones such as TNA (threose nucleic acid) and HNA (hexitol nucleic acid). Structural studies show useful detail about polymerase recognition and backbone geometry, but practical design tools are still limited. The Nucleic Acids Research article on TNA and HNA structural constraints highlights how torsion angle differences and sugar puckering constrain synthesis pathways. This is exactly the kind of problem that benefits from predictive modeling over pure trial and error.

Some sequences fail because the synthesis platform is weak. Others fail because the design ignored how molecules behave before they ever reached the bench.

AI helps most when it narrows experimental uncertainty

AI is useful here when it is applied to specific tasks. The primary gain comes when models predict where a sequence is likely to fail and offer constrained redesigns that keep the biological objective intact.

That can include ranking guide candidates, flagging unstable motifs, modeling variant effects, or prioritizing designs that are easier to manufacture. In translational settings, this often overlaps with broader annotation work. Functional annotation influences what is chosen for synthesis in the first place.

This article on software for biotech gives a practical overview of where tooling is heading. It explains why modeling, data infrastructure, and sequence design software increasingly sit in the same decision stack.

The following video gives a short visual overview of that shift.

What changes when computation is built in early

Three things usually improve when teams integrate computational design before procurement.

• Fewer redesign cycles: Obvious synthesis liabilities are removed before they become ordering failures.
• Cleaner downstream interpretation: If a construct fails functionally, you’re less likely to confuse biology with manufacturing defects.
• Better use of wet-lab time: Scientists spend more effort testing hypotheses and less effort rescuing sequence design mistakes.

Computational design does not eliminate failure. It changes failure from random to informative. That is a large difference in any serious R&D program.

The Future of Synthesis Scale Automation and Biosecurity

The next pressure point for nucleic acid synthesis is whether teams can operate at scale, with enough automation and enough oversight to support therapeutic, industrial and distributed research use.

That pushes synthesis out of the single-project mindset. Once a team needs many constructs, repeated iterations, or production-grade inputs, process architecture matters as much as sequence design.

Scale changes the bottleneck

At larger volumes, manual handoffs become expensive. Every transfer between design, ordering, assembly, QC, and data review creates delay and introduces opportunities for mismatch. Automation therefore matters even when the chemistry is already mature.

The labs that move fastest usually standardize around:

• Structured intake of sequence designs so constraints are captured before ordering
• Automated routing of constructs into the right synthesis and assembly path
• Integrated QC review tied to construct purpose rather than generic acceptance rules
• Traceable data pipelines linking design intent to measured output

This is especially important when one platform supports many use cases at once. A therapeutic RNA template, a CRISPR screening library, and a strain engineering cassette should not be processed under the same assumptions even if they pass through the same infrastructure.

Decentralization raises a real biosecurity problem

Lower-cost benchtop synthesizers create a new challenge. If synthesis capacity shifts away from centrally screened providers, sequence oversight cannot depend on vendor-side review alone.

That concern is now concrete. A frequently unaddressed issue is scalable biosecurity screening for decentralized synthesizers. The Common Mechanism for DNA Synthesis Screening was developed in 2025 with industry experts. The broader need for embedded sequence screening, especially for AI-designed sequences, has become a focus of NIST and HHS in 2025-2026, as discussed in the PMC article on DNA synthesis screening and biosecurity.

The key practical point is simple. Screening has to move closer to the instrument and the workflow.

What responsible implementation looks like

Real biosecurity controls cannot be so heavy that they freeze legitimate research. At the same time, soft policy language without operational tooling does not solve the problem.

A workable model has a few characteristics:

• Embedded sequence checks: Screening happens during design or instrument submission, before synthesis is complete.
• Real-time handling of sequences of concern: The pipeline can flag problematic orders before material is produced.
• Compatibility with AI-generated designs: New sequence generation methods should not bypass safety review because they do not resemble older ordering patterns.
• Auditability: A team should be able to show what was screened, how it was assessed, and what happened next.

The future of nucleic acid synthesis includes faster synthesis and synthesis tied to software that can enforce quality and safety without obstructing iteration.

That combination of automation and computational governance will likely define which synthesis platforms are trusted at scale.

Conclusion Turning Biological Complexity into Actionable Design

Nucleic acid synthesis used to be easy to describe. A team needed DNA or RNA, ordered it, and then focused on the biology. That view no longer matches how serious programs operate.

The synthesis method matters. The workflow matters. QC matters. Sequence design before synthesis matters most because that is where many of the downstream costs and delays are decided. Teams that separate digital design from wet-lab execution usually make their own work harder.

Chemical synthesis remains foundational for many short-format applications. Enzymatic approaches are changing what is practical for longer and more complex constructs. Common workflows such as gene synthesis, PCR assembly, and IVT all bring their own failure modes. Vendor convenience alone does not solve those failure modes.

Computational screening turns this from a reactive process into an engineered one. It helps catch repeats, structure problems, difficult motifs, and context-dependent design issues before they show up as failed assemblies or ambiguous data. It also opens a path into harder design spaces, including non-standard backbones and more automated safety screening for distributed synthesis environments.

For better outcomes, treat nucleic acid synthesis as an integrated discipline. Chemistry or enzymology makes the molecule. Computation makes the outcome more predictable. QC shows whether the molecule received is the one the experiment needs.

That is the shift. The best teams design sequences that are more likely to succeed, and they then verify them in ways that match the real risk of the program.