Heterologous gene expression is widely used across biology and medicine, and often relies on codon optimization to increase protein yields. Here we uncover missplicing as a common and largely unrecognized failure mode of heterologous expression. Using systematically designed libraries comprising over 5,000 synthetic reporter genes and natural human cDNAs, we find that the majority of gene variants expressed in a human cell line are at least partially spliced, and in many variants the spliced isoform dominates, reducing protein output or ablating expression entirely. By analyzing sequence determinants of expression across multiple human cell lines, we uncover a hierarchical architecture of regulatory control, where GC content establishes baseline mRNA levels, local sequence features influence splicing, and tissue-specific codon adaptation to tRNA pools fine-tunes translation efficiency. These findings enable us to develop predictive models of expression and splicing, benchmark current optimization strategies, and design a splice-aware optimization algorithm that substantially improves transgene performance.

Proteins function through a complex interplay of structural and biochemical properties, and mutations can reshape these properties to generate fitness landscapes spanning multiple functional objectives. A central challenge in protein engineering is the need to simultaneously optimize multiple properties. In biocatalysis, for example, practical enzyme development routinely requires the concurrent optimization of catalytic activity, selectivity, stability, and substrate generality. However, despite recent advances in computational protein design and fitness prediction, most existing approaches treat these properties independently and do not explicitly capture the dependencies and trade-offs that govern real-world protein performance. We present SynFit, a multi-objective learning framework that integrates pretrained protein language models with experimental fitness measurements for protein fitness prediction and engineering. SynFit learns both shared and property-specific protein sequence representations through a synergistic contrastive learning strategy, enabling the identification of variants that simultaneously optimize multiple functional properties. Across a large-scale multi-fitness deep mutational scanning benchmark, SynFit consistently outperforms state-of-the-art supervised models trained on individual objectives and more accurately identifies variants that balance competing functional constraints. We further applied SynFit to multi-objective enzyme design for a new-to-nature biocatalytic enantioselective borylation reaction, providing a diverse array of novel cytochrome c sextuple variants in a single round of design with simultaneously improved catalytic activity and enantioselectivity that rival the best variants obtained through directed evolution. Together, these results establish SynFit as a general framework for multidimensional protein fitness prediction and highlight its potential to enable efficient multi-objective optimization in protein engineering, particularly in biocatalysis.

Short-read metagenomic sequencing is widely applied in microbiome research due to its high quality and increasingly more affordable prices. However, it suffers from fragmented reads which limits assembly contiguity and the recovery of complete microbial genomes. In contrast, long-read sequencing, with substantially longer read lengths, can help overcome these limitations. Achieving complete and accurate genome recovery is a central goal in metagenomics. To advance this goal, we present a systematic effort to unify and optimize the long-read sequencing workflow, from experimental sample processing to computational genome assembly, using the CycloneSEQ platform.

Optogenetics integrates living cells and electronics into powerful cell–silicon systems, but prototyping their dynamics remains challenging. Current tools either require robotic liquid transfers into flow cytometers or rely on custom sensors with a narrow dynamic range that limits controller performance. Additionally, current successful optogenetic feedback controllers only operate in chemostats or microfluidic devices that enforce constant growth, because models for growth-aware controller design in batch culture are lacking. Here, we present LEMOS, a low-cost LED-embedded microplate that runs inside a commercial microplate reader. Coupled with a growth-aware multiscale model of gene expression for controller tuning, this platform enables rapid design-build-test-learn cycles for cell-silicon systems. We demonstrate closed-loop set point tracking of gene expression in batch cultures within a standard microplate reader and show how growth dynamics complicate controller selection and tuning. Together, this platform reduces setup overhead and speeds up iteration, enabling accurate real-time optogenetic feedback control.

Our understanding of protein function and evolution is largely based on the relationship between amino acid sequence and overall fold, now effectively captured by computational models. Yet predicting how mutations—shaped by epistasis—alter protein behavior, especially in dynamic or structurally ambiguous regions, remains difficult. Here we present D2D, which combines a self-supervised protein language model with protein-specific evolutionary information to predict mutational effects using little to no task-specific labeled data. D2D captures long-range epistatic interactions, accurately predicts single and higher-order mutation effects on protein thermostability and binding, without being trained on the task. When fine-tuned, D2D outperforms state-of-the-art methods on latent driver cancer mutations and co-occurring proliferation-enhancing mutations across independent experimental studies. Unlike most existing approaches, D2D avoids biases linked to solvent accessibility or to multiple sequence alignment depth and quality, making it particularly effective for disordered or surface binding regions where structure-based predictors typically falter. Overall, D2D provides a general framework for modeling mutational effects in proteins with limited experimental or structural information.

How microbial communities maintain robust and reproducible ecological functions despite their extraordinary taxonomic diversity remains an open question. Here we show that functional organization in microbial communities can be uncovered by repurposing Principal Component Analysis to focus on directions of lowest variance in taxon abundance data, rather than maximal variance. These least-variance components are statistically significant and correspond to ecological constraints on taxon abundances that are consistently fulfilled across samples. Using consumer-resource models, we show that these constraints arise from resource-mediated interactions and express biomass conservation, effectively grouping taxa into producer and consumer guilds. We validate this interpretation in simulated communities and experimental systems under competition and cross-feeding. Finally, we show that low-variance structure is ubiquitous in natural microbial communities and reveals a sparse network of taxa with disproportionate influence on community structure. Together, our results establish low-variance components as indicators of ecological constraints linking taxonomic diversity to functional organization.

Noncanonical amino acid (ncAA) incorporation at the protein N-terminus provides a powerful strategy for installing defined chemical handles while minimizing perturbation of internal protein sequences. However, highly efficient initiation-based ncAA incorporation systems suppress the native methionine pathway by removing methionine or methionyl-tRNA synthetase, limiting their use for proteins containing internal methionine residues. Here, we developed an orthogonal initiation system for selective N-terminal ncAA incorporation into proteins in intact cell-free translation systems. We systematically profiled background initiation from all 64 codons in reconstituted translation systems and identified low-background artificial initiation codons. Engineered initiator tRNAs, termed tRNAIniTx, were then designed to decode selected codons and support ncAA-dependent initiation. The optimized CAC/tRNAIniTx04GUG pair enabled efficient N-terminal incorporation of N-biotinyl-L-phenylalanine without removing methionine or methionyl-tRNA synthetase, reaching over 90% incorporation. The system was further extended to p-azido-L-phenylalanine and to an E. coli extract-based cell-free translation system. Finally, N-terminally biotinylated proteins were directly immobilized on streptavidin biosensors for purification-free biolayer interferometry analysis of computationally designed Brd4BD2 binders. This work establishes a codon-guided orthogonal initiation strategy for N-terminal protein functionalization while preserving the native methionine translation pathway.

Reaction-diffusion circuits generate self-organized spatial patterns through local activation and long-range inhibition, but synthetic implementations in mammalian cells have been limited by the differential-diffusion requirement. Here, we introduce a novel architecture, juxtacrine activation with paracrine inhibition (JAPI), where the activator propagates through cell-cell contacts rather than diffusion. We demonstrate mathematically and numerically that JAPI accesses the same patterning regimes as classical diffusion-based circuits with one fewer free parameter. We then engineer compact synNotch-based JAPI circuits in mammalian fibroblasts and demonstrate their sufficiency for self-organized patterning through tunable, size-limited signal propagation. Functionalized to spatially control morphogen secretion, these circuits perturb feather bud formation on adjacent embryonic chicken epidermis. Finally, we develop a library-based approach to explore coupled, dual-JAPI circuits with tunable cross-inhibition, enabling programmable interactions between patterns and access to a broad morphospace of spatial states. Together, JAPI provides a compact, modular platform for programming self-organized multicellular patterning.

Biological nitrogen fixation in some filamentous cyanobacteria occurs in heterocysts, yet the subcellular organization of nitrogenase in these cells remains poorly defined. Using fluorescence microscopy of living Anabaena sp. PCC 7120, we found that a NifH-GFP fusion forms discrete puncta restricted to mature heterocysts, whereas constitutively expressed GFP in vegetative cells and a nifB-linked GFP reporter remained diffuse. To contextualize this phenotype, we built a homology-aware cyanobacterial condensate-prioritization framework across 31,028 proteins from seven proteomes, integrating ESM-2 embeddings, 27 biophysical features, and 44 curated condensate-associated positives. Because direct positives remain limited, we present the model as a ranking resource rather than a calibrated classifier. Under a deliberately conservative nitrogenase-withheld analysis that removed nitrogenase-family labels and homologous clusters from the training data, NifH retained a median rank in the highest-scoring 4.8% of cyanobacterial proteins. catGRANULE 2.0 also assigned high scores to nitrogenase iron proteins, while having only modest rank concordance to the whole atlas. Together, these data identify heterocyst-restricted NifH-GFP puncta as evidence of sub-cellular spatial organization and provide a curated resource for prioritizing cyanobacterial condensate candidates, some of which are also important for nitrogen fixation.

Tandem gene duplication drives antibiotic resistance, metabolic adaptation, and gene-family expansion in bacteria, but no tool detects them in reference genomes, discovers their junctions in isolate sequencing, and quantifies the junctions in population samples. Existing callers (e.g. breseq) detect duplications without classifying formation mechanisms and often fail to quantify the duplication.  Tandem has 3 modules. Module 1 detects reference-genome duplications by NUCmer self-alignment and classifies each by homologous-recombination signature and the junction microhomology length. Module 2 confirms junctions in whole-genome sequencing at user-nominated coordinates after user inspecting the coverage plot. Module 3 quantifies known junction in population sequencing using the novel Junction Read Ratio (JRR). On 280 artificial population tests across seven bacterial species, Tandem achieves 100% recall and 4.3% mean absolute error. Applied to experimentally evolved Pseudomonas fluorescens SBW25 populations, Tandem resolves multiple co-segregating duplication fragments.

Root competence, the ability of soil bacteria to establish and grow on plant roots, is a key ecological trait influencing plant nutrition, growth, and health. However, identifying genomic determinants of root competence across bacteria remains challenging, in part because model generalizability depends strongly on how genomes are represented. Traditional approaches based on curated annotations are incomplete and biased toward well-characterised organisms and functions, limiting generalization. Sequence-similarity clustering improves coverage but yields high-dimensional features relative to dataset size, hindering training. Foundation models offer an alternative by learning compact representations without relying on prior annotation. Here, we compared pretrained genome representations from protein and DNA foundation models (ESM-2, Bacformer, DNABERT-S) with annotation- and clustering-based features (KEGG orthology, OrthoFinder protein families) for predicting root competence using synthetic microbial community data from Arabidopsis thaliana and assessed generalisability across bacteria. When training and test sets contained taxonomically related bacteria, most approaches performed similarly. However, when test bacteria belonged to phyla entirely absent from training, reflecting high evolutionary separation across all levels of bacterial classification, only pretrained protein representations retained predictive performance. Bacformer-derived representations, which incorporate genomic context, supported the strongest generalisation, suggesting that conserved genomic organisation contributes to predicting root competence. Feature attribution quantifying protein contributions to model decisions linked root competence to TonB/SusD-dependent receptors, small-molecule transporters, and unannotated proteins with conserved regulatory motifs and homology to carbon starvation-response loci. Protein foundation models support generalisation across evolutionarily distant bacteria and identify genomic determinants of root competence, including unannotated proteins.