bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2025.10.24.684421

MillionFull enables massive, full-length enzyme sequence-fitness data collection at low cost for machine learning-guided enzyme engineering

Li, J.·Erichsen, B.·Krarup, S. R.·Yuan, S.·Jijakli, K.·Karst, S.·Yang, L.·Nielsen, A. T.
View original source ↗Export metadata ↓

Abstract

Machine learning holds great promise for accelerating enzyme optimization, but its power is fundamentally constrained by the limited availability of sequence-fitness data. Here, we introduce MillionFull, a low-cost method that enables high-throughput full-length sequence- fitness mapping for enzymes of arbitrary length. Each run yields on the order of 10-10 data points, capturing sequence-function relationships at unprecedented scale. By overcoming the data bottleneck, MillionFull provides a foundation for dramatically advancing AI-driven enzyme engineering.

Explore related subjects

synthetic biologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Li, J., Erichsen, B., Krarup, S. R., Yuan, S., Jijakli, K., Karst, S., Yang, L., Nielsen, A. T.. 2025-10-25. MillionFull enables massive, full-length enzyme sequence-fitness data collection at low cost for machine learning-guided enzyme engineering. https://doi.org/10.1101/2025.10.24.684421

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Machine learning guided design of strong transcriptional activation domains

Artificial intelligence has accelerated the design of structured proteins, but it remains difficult to design functional intrinsically disordered proteins because of weak structural constraints and poorly defined sequence-to-function relationships1. Activation domains (ADs), transcription factor regions that recruit coactivators, are an excellent test case for the de novo design of intrinsically disordered sequences. ADs are one of the oldest groups of functional intrinsically disordered sequences2 and designed ADs have applications in synthetic biology, biotechnology, and agriculture. We computationally generated 500,000 candidate sequences that resembled known ADs and predicted AD function with extant machine learning models, including three neural networks. From high-scoring sequences, we selected 18,750 for experimental validation in yeast. Surprisingly, >90% of the 16,770 well-measured sequences function as ADs. Sequences predicted to be active by multiple models were stronger, motivating development of a fast and accurate consensus AD predictor, MADRIGAL. A number of our designed ADs retain functionality across fungi, plants, and human cells. This work illustrates a generalizable paradigm for incorporating sequence-to-function models into the design loop to efficiently generate functional intrinsically disordered sequences.

synthetic biology↗
bioRxiv2026

E. coli-produced nanobody fusions as a pan-serotype platform for AAV purification, quantification, and detection

Research, production and application of the recombinant adeno-associated virus (rAAV) require versatile tools for rapid and cost-effective advances. We developed fusion constructs with the pan-serotype, AAV-binding VHH-antibody-fragment named nanobody AAVX and devised procedures for AAV purification, quantification, and detection. Starting from an amino acid sequence, we cytosolically expressed in E. coli nanobody AAVX carrying various tags and fusion partners. Very efficient soluble production followed by a one-step immobilized metal-ion affinity chromatography (IMAC) yielded mostly >90% pure protein. Crosslinking AAVX with an agarose matrix enabled efficient purification of intact AAV2 and AAV5 capsids from cell lysates, as assessed by qPCR and ELISA. AAVX also supported effective AAV quantification by ELISA, as capture reagent and, when fused to luciferase, as detection reagent. Using tagged AAVX and streptavidin biosensors, we developed a rapid and label-free quantification assay of AAV capsids by biolayer interferometry (BLI). Finally, immunostaining with an AAVX-eGFP fusion visualized AAV capsids in fixed HEK-293 cells by fluorescence microscopy. Collectively, these results establish nanobody AAVX fusion proteins as versatile and readily customizable tools for AAV research and bioprocessing.

synthetic biology↗
bioRxiv2026

Duality Vaccination through Nanodesign of Click-Ligated B-Cell Inhibitory Receptor Bridging

Synthesis of precise and robust molecular architectures is critical to harness the power of the immune system. The dimensions, rigidity and multivalency of nanoparticles contribute to eliciting high levels of antibodies from vaccination. However, antibodies vary hugely in quality: some antibodies are highly effective at neutralizing pathogens, while others provide little or no benefit. Here we establish switchable and modular platforms for protein and nanoparticle assembly that program the target of antibody responses. We introduce Duality vaccination: the simultaneous delivery of activating and inhibitory stimuli for focusing immune responses. To construct activatory immunogens, we employ click biology with spontaneous amidation to functionalize plug-and-display self-assembling porous dodecahedral nanocages. We construct inhibitory immunogens using SpyMask protease-gated bispecific assembly, ligating the antigen to binders that bridge the B cell receptor to immune checkpoint receptors. This dual immunogen architecture tailors the elicited antibodies to down-modulate undesired B cell responses. We establish the effect of temporal and spatial separation of activatory and inhibitory immunogen delivery. We further explore sensitivity to valency, linker flexibility and adjuvant delivery, employing model targets and coronavirus antigens. Dual programmable functionalization identifies key nanodesign features to control in vivo outputs, providing logic-gating towards urgent challenges in the creation of protective vaccines.

synthetic biology↗