Building Synthetic Cells Together Requires a Common Standard of Evidence
RESEARCH
2026-11-17
Perspective for SynCell Asia

The challenge of building a cell together

Constructing a synthetic cell across multiple laboratories requires assembling independently developed biological modules into a coherent, functioning system. SynCell Asia has proposed distributing module development among specialized groups and integrating those components through a centralized, AI-driven biofoundry [1]. That distributed model, however, imposes a strict scientific requirement: experimental evidence generated in one laboratory must be directly usable by another. Sharing physical constructs and reagents is insufficient unless teams also share the quantitative knowledge needed to make them work together.

Consider coupling a membrane transport module built in one laboratory to a gene expression system developed in another. Whether the two modules function together depends on how substrate influx, energy demand, and local reaction conditions evolve over time. Endpoint summaries such as a final protein yield or an apparent permeability coefficient do not establish whether those dynamic requirements match. Predicting how the coupled system will behave requires knowing specifically which biological function was measured under which conditions, and how those measurements constrain quantitative expectations for the combined assembly.

Here, we argue that standardized functional evidence should serve as core scientific infrastructure for building synthetic cells together. Explicit standards record the biological configuration and assay conditions behind each dataset, while quality control (QC) and calibration make observations interpretable and comparable across groups. Automated biofoundries can generate such records reproducibly at scale, enabling structured experiments and quantitative models to extract functional relationships that guide subsequent design cycles and define when those relationships transfer across laboratories.

Define the biological meaning of a measurement

A functional value has clear scientific meaning only when context, environment, and measurement are separated explicitly. In our framework, context denotes the biological configuration specified by the design itself, encompassing the chosen host or cell-free reaction architecture together with the participating molecular components and how they are arranged and coupled. Because context follows directly from design logic, independent laboratories can in principle implement it consistently. Environment, by contrast, refers to the operating regime imposed for a given task, including variables such as temperature, pH, and substrate supply, which can likewise be held constant for a defined application. Measurement sits outside the biological system altogether, covering the assay chemistry, instrumentation, calibration protocols, and data-processing pipelines used to observe and quantify behavior.

Recent work by Wang and colleagues illustrates why decomposing mixed biological signals matters in practice [2]. By separating host-dependent background expression from the intrinsic activity of a paired K1.5 promoter–RNA polymerase system, the authors resolved a quantitative sequence–activity landscape that guided the de novo design and experimental validation of insulated promoters. Stripping out confounding host activity exposed sequence-to-function relationships that composite expression readouts had previously obscured. Within the tested system, isolating the target activity supported both mechanistic interpretation and predictive design, offering far stronger leverage for discovery than simply improving correlation scores on an unseparated signal.

A shared functional record should therefore tie each sequence-verified molecular system to its design context, operating environment, measured quantity, experimental controls, raw observations, and versioned analysis pipeline, complete with uncertainty estimates [3]. Both host-associated contributions and derived intrinsic parameters belong in that archive. Depending on the question at hand, a context effect might be subtracted out during parameter estimation, modeled explicitly as an interaction term, or studied as a biological phenomenon in its own right, allowing investigators to resolve both intrinsic module behavior and its dependence on surrounding biological conditions.

Build quality control and calibration into the workflow

Quality control must judge whether an observation is valid for its stated scientific purpose without erasing genuine biological variation. Technical failures, including construct mismatches, mislabeled samples, and failed experimental controls, require explicit diagnosis and filtering. Reproducible shifts driven by resource competition or by changes in host growth state and inter-module coupling, however, often reflect how the biological system actually behaves. Data systems should therefore retain raw records alongside documented exclusion criteria, preserving unresolved discrepancies rather than silently dropping outliers. Agreement with a prior model can never serve as the criterion for accepting or rejecting a measurement.

At the measurement layer, Relative Promoter Units (RPUs) showed that a shared reference standard can align transcriptional activity measurements across instruments and laboratories [4]. We propose extending that principle by pairing routine QC with multipoint calibration instead of single-point normalization. Calibration series should span the intended dynamic range, using blanks, replicates, and independent check standards to characterize background floors and detector saturation, test for departures from the assumed response curve, and track instrumental drift through repeated reference reads. Every resulting unit conversion should report explicit uncertainty bounds together with the operating conditions under which the calibration holds.

Instrument calibration alone suffices for cross-laboratory comparison only when design context and operating environment are matched. Changing either the design logic or the operating conditions introduces biological transfer effects that require empirical testing. Paired measurements of shared reference constructs or modules across defined changes let investigators distinguish simple proportional scaling from nonlinear transfer or rank-order inversions among candidates. Empirical transfer maps should anchor multiple reference constructs across the relevant functional range and report uncertainty intervals, a validated operating domain, and test performance on held-out constructs or conditions. Where an empirical mapping breaks down, that failure is diagnostic, exposing unmodeled mechanistic dependencies and defining regimes in which direct remeasurement remains indispensable.

Automated biofoundries can build this integrated QC and calibration into everyday execution. Existing workflow abstraction hierarchies already map high-level experimental designs onto executable laboratory operations [5]. In practice, construct verification, reference panels, raw-data capture, QC decisions, and calibration versions should travel together as one traceable record. At the software layer, large language model (LLM) agents can structure experimental metadata against community standards and run deterministic validation checks before coordinating repository deposition and cross-database exchange. Under explicit, versioned validation rules that route unresolved conflicts to human review, high-throughput experiments accumulate into auditable records that other laboratories can directly interpret and extend.

Use structured evidence to discover design rules

Discovering transferable design rules requires experimental layouts structured to separate competing mechanistic explanations. Paired assays test quantitative transfer across conditions, and controlled combinatorial designs expose interactions between modules, yet time-resolved measurements are often decisive because distinct kinetic regimes can reach the same endpoint. Two reactions with identical final protein yields, for example, may reflect either a rapid early burst of synthesis or slower, sustained production. Those two trajectories carry very different consequences when gene expression must be coupled to transmembrane transport or energy regeneration. Recording complete time courses keeps these dynamical distinctions available for mechanistic testing.

What generalizes from such experiments is a set of conditional quantitative relationships linking inputs, outputs, resource demands, and module interactions within explicitly defined operating boundaries. In membrane–expression integration, such rules relate substrate delivery to internal consumption and protein synthesis over time under compatible reaction conditions. Community databases should pair each candidate design rule with both supporting and conflicting observations, parameter uncertainties, and results from independent validation tests. Documenting failure boundaries alongside validated regimes gives teams concrete quantitative criteria when selecting modules for a new integration task.

Treating measurements as structured evidence also changes how investigators choose the next experiment. Beyond empirical screening for higher output, subsequent assays can be selected specifically to separate competing mechanistic hypotheses. In a stalled membrane–expression reaction, for instance, researchers can formalize rival models centered on substrate-transport limitation versus internal energy depletion and apply experimental perturbations under which the two models predict maximally divergent responses. New observations then refine both the quantitative model and its domain of validity, sharpening mechanistic understanding alongside construct performance.

Build a shared basis for synthetic-cell collaboration

As an initial practical test of this framework, we are connecting the Polaris Kit with the cell-free transcription–translation (TXTL) Synthetic Cell track of the SynBio Challenges. Building on recent community calls for biofoundry data interoperability [6], this joint pilot couples physical toolkit standardization with distributed functional testing. Polaris provides a unified framework for hierarchical DNA assembly, genomic integration, and quantitative characterization in Escherichia coli and Saccharomyces cerevisiae [7], whereas the competition contributes an independently designed regulatory-sequence challenge assayed in an E. coli-lysate TXTL system [8]. Across this pilot, standardized construct definitions are linked with explicit records of design context, operating environment, and measurement protocol alongside structured QC logs and agent-assisted curation. Extending beyond the published cellular toolkit [7], prototype agent-assisted modules for QC checks, data ingestion, and cross-database exchange are also being integrated to support evaluation across participating teams. Because cellular assays and cell-free TXTL reactions differ in both biological configuration and operating environment, quantitative transfer between the two formats cannot be assumed and must instead be calibrated empirically using shared reference constructs measured in both systems. The decisive benchmark for this pilot is straightforward: whether functional records generated by different participating teams can be independently audited and reused quantitatively in subsequent rounds of design.

The shared resource that synthetic-cell consortia ultimately need is a body of quantitatively tested knowledge explaining how biological functions work together. Explicit definitions of context, environment, and measurement make differences across experiments interpretable, while built-in QC and multipoint calibration allow observations from separate groups to be combined with known uncertainty. Structured experiments and quantitative models can then distill conditional design rules from those shared records. Standardization succeeds when rules established in one laboratory measurably sharpen mechanistic prediction and construct design in the next.

References

1. Dai, Z., Aoki, W., Chaiyen, P. et al. A framework for building a synthetic cell from the SynCell Asia Initiative. Nature Biotechnology 44, 1084–1088 (2026). https://doi.org/10.1038/s41587-026-03153-w.

2. Wang, H., Xiang, Y., Liu, Z. et al. De novo design of insulated cis-regulatory elements based on deep learning-predicted fitness landscape. Nucleic Acids Research 53(12), gkaf611 (2025). https://doi.org/10.1093/nar/gkaf611.

3. Vitalis, C., Vidal, G., Samineni, S. P., Fontanarrosa, P. & Myers, C. J. A framework for a standard-enabled FAIR data management workflow for synthetic biology. ACS Synthetic Biology 15(1), 1–8 (2026). https://doi.org/10.1021/acssynbio.5c00813.

4. Kelly, J. R., Rubin, A. J., Davis, J. H. et al. Measuring the activity of BioBrick promoters using an in vivo reference standard. Journal of Biological Engineering 3, 4 (2009). https://doi.org/10.1186/1754-1611-3-4.

5. Kim, H., Hillson, N. J., Cho, B.-K. et al. Abstraction hierarchy to define biofoundry workflows and operations for interoperable synthetic biology research and applications. Nature Communications 16, 6056 (2025). https://doi.org/10.1038/s41467-025-61263-6.

6. Judd, J. Biofoundry data interoperability requires community action. ACS Synthetic Biology (published online 20 October 2026). https://doi.org/10.1021/acssynbio.6c00469.

7. Xie, R., Wang, R., Zhan, Y. et al. Polaris: A toolkit for standardized construction, integration, and quantitative testing in Escherichia coli and Saccharomyces cerevisiae. Synthetic and Systems Biotechnology 20, 32–45 (2027). https://doi.org/10.1016/j.synbio.2026.10.012.

8. SynCell Asia Initiative. SynBio Challenges: Synthetic Cell track and TXTL-based sequence design challenge (2025 competition information). https://isynbio.siat.ac.cn/cxqjzx/syncell_asia/resources/2025101410381476858/index.html