Synthetic biology has advanced quickly in the last decade, and a lot of that is because of next-generation sequencing (NGS). In synthetic biology, engineering principles are applied to biological systems, to modify or create new biological organisms or molecules. Because of this, the working paradigm is the design-build-test-learn (DBTL) cycle.
With the instructions to every organism, system, or molecule being encoded in DNA, NGS is crucial for all the DBTL steps. Additionally, NGS has become cost-effective enough to enable library-scale characterization, making it possible to sequence not just a few clones but entire construct libraries spanning millions of variants.1
In this article, we’ll focus on how NGS advances support the DBTL cycle, with special emphasis on the “design” and “build” phases.
Mining sequence data to inform design
The design phase increasingly relies on large-scale genomic and metagenomic datasets as source material for identifying novel genes, enzymes, or pathways. Single-cell genomics technologies work well for eukaryotic cells but not for microbial cells,2 which are a big part of synthetic biology. However, some researchers and companies have cracked that wall already.
“Single-cell technology itself is nothing new,” Ahir Pushpanath, Ph.D., Director of Biocatalysis at bitBiome, explains. “But we've cracked how to do it from microbial samples, isolating individual microbial cells into hydrogel capsules, where lysis and whole-genome amplification occur within the same compartment. This is the bit-MAP technology.” This allows bitBiome to sequence individual microbial genomes at scale, accessing genetic diversity from unculturable organisms that shotgun metagenomics frequently misses.
According to Dr. Pushpanath, single-cell sequencing makes a difference. “With shotgun metagenomics, short reads are first assembled into contigs, and then binning is used to group those contigs into draft genomes. This can be challenging. Strain variation, uneven coverage, and low-abundance organisms can lead to fragmented genomes or misassigned contigs, which can reduce confidence in linking genes back to specific organisms.”
From all the data collected, bitBiome created a database. “We’ve built a proprietary microbial sequence database containing roughly 2.5 billion genes,” reports Dr. Pushpanath. These NGS-generated databases serve as parts catalogs for synthetic biology, providing candidate sequences for enzyme engineering, pathway construction, and directed evolution campaigns.
Regardless of whether metagenomics or single-cell sequencing is used, NGS is key to generate data to have a strong starting point and knowledge for the build phase of the cycle.
QC to verify what was built
Once constructs are designed, the build phase requires rigorous QC to confirm that what was synthesized matches what was intended. NGS (including long-read sequencing) is needed to verify full construct integrity, as well as detect insertions, deletions, frame-shifts, and unwanted sequence motifs across entire libraries rather than spot-checking a handful of clones.
One of the companies building the largest DNA constructs is Ansa Biotechnologies. “We're delivering up to 50 kb or 65 kb now,” CEO, Jason T. Gammack, notes. “That requires a whole different level of QC. We do de novo sequencing to the entire plasmid and sequence at a deep depth. And we count molecules and report to the customer the actual percentage of full-length perfect molecules within the tube.”
For complex or long DNA constructs, long-read sequencing is especially important because short reads cannot span repetitive regions, secondary structures, or full-length assemblies. Platforms like PacBio HiFi and Oxford Nanopore can generate long reads with high accuracy, enabling end-to-end verification of multi-kilobase constructs Catching errors at the build stage prevents wasted resources on non-functional constructs downstream.
For Gammack, the positive outcome of all this is speed. NGS of larger constructs allows scientists to know exactly what they have in the tube, so they spend less time assembling fragments and testing quality. “The QC data that we deliver to customers could be used in any type of regulatory submission,” Gammack explains. This is definitely important for any work on commercial products.
High-throughput screening powered by NGS
In the “test” phase, NGS can track millions of genetic variants through functional screens, connecting sequence (genotype) to performance (phenotype). “NGS remains essential because modern enzyme and strain engineering generates very large variant libraries, and sequencing lets us know which variants perform best,” Dr. Pushpanath says. Deep sequencing of variant libraries after selection rounds provides quantitative enrichment data, revealing which sequences are best, and how experimental parameters affect outcomes.3
NGS enables discovery of rare functional variants even at very low frequencies within large libraries, something that colony-picking and Sanger sequencing simply cannot achieve at the same scale. Applications range from directed evolution or pathway optimization, to combinatorial library screening across both microbial and cell-free systems.
Feeding data back into the next cycle
The “learn” phase is where researchers use the test data (including NGS) to guide the next DBTL cycle toward the desired outcomes. Bioinformatics and statistics can help learn more for the next DBTL cycle. But with the quantity of data from NGS, machine learning and artificial intelligence (AI) can go beyond other tools to improve the next design iteration.4
Ansa approaches this head on. “We gather DNA synthesizer performance and NGS data, and use machine learning to optimize DNA synthesis,” Gammack says. This way, Ansa can build larger and larger fragments, as well as those of higher complexity.
“For me, the biggest shift will be the move toward predictive biological design driven by AI,” notes Dr. Pushpanath. He explains that once sequence, phenotype, and performance metrics started to be combined, and enough data obtained, “we can start understanding how to design life.”
And while that might be years away, the learn phase benefits heavily from the additional NGS data that comes from advances allowing single-cell genomics, deeper sequencing, epigenetic modifications detection, and more.
What’s next for NGS in synthetic biology
Long-read sequencing is becoming standard for validation of complex constructs. As accuracy improves and costs decline, routine verification of multi-kilobase synthetic constructs will become the default rather than the exception.
Declining NGS costs are democratizing library-scale QC. All platforms become cheaper and produce more data, which means that verification is accessible to academic labs and startups, not just bigger companies.
AI and machine learning models trained on NGS-generated datasets might allow us to skip or predict many DBTL cycles, closing the gap between computational design and experimental validation.5
Most NGS developments improve synthetic biology by providing researchers with more tools and data to design, build, test, and learn. Synthetic biology developments feed back into NGS too, with DNA polymerases, ligases, retrotranscriptases, and other NGS-relevant enzymes being created, improved, and engineered.
NGS and synthetic biology will continue to co-evolve, one providing data, the other increasing the complexity of outputs, each shaping what the other can even attempt.
References
1. Rai K, O’Connell RW, Piepergerdes TC et al. Ultra high throughput mapping of genetic design space. Nature. 2026. doi:10.1038/s41586 025 09933 9
2. Hosokawa M, Nishikawa Y. Tools for microbial single cell genomics for obtaining uncultured microbial genomes. Biophys Rev. 2024;16:69 77. doi:10.1007/s12551 023 01124 y
3. Ito T, Nguyen TD, Saito Y, Kurumida Y et al. Selection of target binding proteins from the information of weakly enriched phage display libraries by deep sequencing and machine learning. mAbs. 2023;15(1). doi:10.1080/19420862.2023.2168470
4. Rai K, Wang Y, O’Connell RW et al. Using machine learning to enhance and accelerate synthetic biology. Curr Opin Biomed Eng. 2024;31:100553. doi:10.1016/j.cobme.2024.100553
5. Groff Vindman CS, Trump BD, Cummings CL et al. The convergence of AI and synthetic biology: the looming deluge. npj Biomed Innovations. 2025;2:20. doi:10.1038/s44385 025 00021 1
About the author: Darío Sánchez Martín is a scientific writer and molecular biologist, specializing in biotechnology, molecular biology, and nanotechnology. He holds a Ph.D. in Biotechnology from Uppsala University, where he developed nanoparticle-based visual and magnetic assays to detect antimicrobial resistance genes. He co-founded and works at Helixa Communications as a scientific writer, helping life-science companies turn complex science into clear and persuasive content for diverse audiences.