iiinova — where your digital self keeps evolving

We're still pushing DS on length and scale with more in the pipeline. In the meantime, check the preprint for lots more details on DS-Gold: www.biorxiv.org/content/10.6...

biorxiv.org

010

Many GGA library assembly methods are limited by the number of orthogonal junctions that can coexist in a shared reaction. DropSynth-Gold performs each assembly in its own droplet. Library scale is determined primarily by barcode diversity, not junction diversity. 5/n

We benchmarked 6 libraries spanning 5-12 fragments per gene corresponding to lengths of ~1 kb to ~3 kb. GGA had comparable coverage and fidelity with a slight hit on uniformity. Fidelity (1 in ~1200 bp) remained stable across all lengths. 4/n

Gel image with assembly bands from 1 kbp to 3 kbp.

100

We kept everything that makes DropSynth scalable: barcoded bead capture, emulsion compartmentalization, and pooled processing, but replaced PCA with Type IIS Golden Gate Assembly inside droplets. 3/n

100

Traditional DropSynth uses PCA, which becomes increasingly prone to cross-hybridization artifacts, partial assemblies, and amplification bias especially as fragment count increases. 2/n

100

Alison Tebo @agtebo.bsky.social · 17/02/2026

We just preprinted DropSynth-Gold where we replaced PCA with Golden Gate Assembly (GGA) inside emulsion droplets, extending multiplexed gene synthesis to ~3 kb constructs assembled from up to 12 fragments while preserving the scalability of DropSynth. 1/n

2178

Reposted by Calin Plesa

We’re excited to announce the next Fluorescent Proteins & Biological Sensors conference! Abstract submissions are now open—hope to see you there. www.janelia.org/you-janelia/...

janelia.org

Fluorescent Proteins and Biological Sensors VIII

Organizers Eric Schreiter, HHMI/Janelia Research Campus Alison Tebo, HHMI/Janelia Research Campus Invited Participants Xiaolu Cambronne, University of Texas at Austin Katharine Diehl, University of

02014

Reposted by Calin Plesa

Addgene @addgene.bsky.social · 04/03/2026

Just released! Fluorescent Protein Libraries from the Plesa Lab. Benchmarking protein expression, developing imaging tools, fluorescent protein engineering, and more. www.addgene.org/pooled-libra...

addgene.org

Addgene: Fluorescent Protein Libraries

Use to benchmark protein expression, develop imaging tools, or to serve as a starting point for fluorescent protein engineering.

0146

To enable reuse by the research community we're making the two parental FPBase libraries available on @addgene.bsky.social as pooled plasmid libraries (Addgene #245482 #245483). Use them. Shuffle them. Train on them! addgene.org/Calin_Plesa/ 12/n

043

So if you want ML to explore distant optima, don’t just improve the model. Engineer the training distribution. Create diversity experimentally. Validate it functionally. Then let the model interpolate across it. 11/n

120

ML-generated functional proteins show more mosaic structure than even the shuffled library. The model learns to recombine sequence segments across parental families in new ways (remixing at scale) 10/n

100

Diversity metrics (clustering, k-mers, nearest-neighbor identity, mosaic structure, embedding geometry) show: 1) DNA shuffling expands dispersion 2) FACS constrains but preserves breadth 3) ML redistributes occupancy across new regions 9/n

100

We synthesized and screened them with hundreds of ML-generated proteins showing reproducible blue fluorescence. Some extend far beyond known natural clusters, including variants with <30% identity to any @fpbase.org protein. 8/n

120

We FACS sorted for blue fluorescence creating a high confidence 7,812 seq BFP training set. We fine tuned ProtGPT2 on this and generated 11,000 de novo sequences which were pruned for max diversity to set of 1,500 for synthesis. 7/n

100

Even after heavy recombination, the β-barrel scaffold proved surprisingly robust. The shuffled library retained substantial fluorescence while increasing diversity 3x relative to parents. We expanded the manifold without collapsing function. 6/n

We next generated large numbers of new chimeric variants, that bridge distant homologs by DNA shuffling across the entire parental set. This synthetic recombination expanded sequence dispersion. 5/n

100

Using β-barrel fluorescent proteins (FPs) as a model, we synthesized a large fraction of the known β-barrel FPs from @fpbase.org (620 seqs), in two codon versions using DropSynth resulting in two large parental libraries spanning natural diversity. 4/n

We hypothesized that by experimentally expanding sequence diversity, we could convert extrapolation into interpolation by expanding the known manifold. 3/n

100

PLMs do best when interpolating within known sequence space. But many protein families are sparsely sampled. Also global fitness optima may reside in distant regions of sequence space far from well studied seqs. So many design problems are extrapolative. 2/n

100

Calin Plesa @calin.bsky.social · 16/09/2025

New lab preprint: if ML struggles with extrapolation, let's expand the diversity of training data with gene synth, DNA shuffling, and ML gen.... also lots of #FluorescentProteins #ProteinEngineering #MachineLearning #SyntheticBiology 1/n

2135

This year's UO iGEM team is building a histamine-responsive probiotic to help tame mast cell flares in MCAD. Help us get more students to the iGEM Jamboree in Paris this October (gifts via UO Foundation may be tax-deductible): duckfunder.uoregon.edu/project/46947

duckfunder.uoregon.edu

iGEM 2025

Help University of Oregon raise $6,000 for the project: iGEM 2025. Your gift will make a difference!

040

The BioE dept in the Knight Campus is a great place to work and is unique in many ways! We are expanding over the next few years with our second building opening in 2026. Come join us, applications begin review on October 15. ...4/n

020

Computational & Data-Science Neuroengineering: neural signal processing and analytics, closed-loop behavior tracking, neuromorphic computing and brain-inspired AI, spatial multi-omics analysis ...3/n

121

Experimental & Translational Neuroengineering: neural interfaces, neurophotonics for brain stimulation/recording/imaging, pre-clinical bioelectric medicine or brain injury repair models, brain organoid technology, disease models ...2/n

Cees Dekker @ceesdekker.bsky.social · 13/08/2025

Our BioE dept at @uoknightcampus.bsky.social has an open rank search for two complementary tenure-track faculty positions in Neuroengineering focused on: 1) Experimental & Translational Neuroengineering and 2) Computational & Data-Science Neuroengineering. careers.uoregon.edu/en-us/job/53... ...1/n

243

Reposted by Calin Plesa

WORK! I currently have 5 (!) open positions for postdocs/PhD students in our CDlab for projects: - Nanopore protein sequencing - Archaeal CDV cell division - Microfluidics for synthetic cells - Nuclear Pore Complex - Origami mimics of peroxisomes Please apply! ceesdekkerlab.nl/come-join-us/ RT=👍

22330

Reposted by Calin Plesa

Karl Romanowicz @kroman.bsky.social · 16/08/2025

Our new Science Advances paper is out! 🎉 The Plesa Lab's first, it builds on DropSynth technology as a proof of concept for large-scale synthetic gene libraries, showcasing a synthetic metagenomics approach to studying antibiotic resistance at scale. 👉 www.science.org/doi/10.1126/...

science.org

Exploring antibiotic resistance in diverse homologs of the dihydrofolate reductase protein family through broad mutational scanning

DropSynth technology enables scalable and cost-effective exploration of antibiotic resistance across the DHFR protein family.

0126

Reposted by Calin Plesa

Nozomu Yachie @nzmyachie.bsky.social · 20/06/2025

CAGT was fun! Thanks @carldeboer.bsky.social @sudpinglay.bsky.social and the de Boer lab for organizing! Folks from Seattle, Oregon, and other places. Great community. Arman gave a usual super talk, and Sanchit and Dayag won poster prizes☺️

093

Calin Plesa @calin.bsky.social · 29/05/2025

Amazing PhD defense by Dr. Natanya Villegas! She pioneered CRISPR-Cas9 and RNA work in our lab and drove multiple tech-dev projects. She is looking for opportunities in the PNW so reach out -> @trienetoscience.bsky.social

090

Calin Plesa @calin.bsky.social · 28/05/2025

Huge congrats to Dr. Andrew Holston on a successful PhD defense!! Andrew has been working on the large-scale characterization and engineering of chimeric receptor histidine kinases. He’s looking for roles in biotech/academia, let’s connect if you’re hiring! -> @hkalltheway.bsky.social

1152

Reposted by Calin Plesa

The Knight Campus at the University of Oregon @uoknightcampus.bsky.social · 27/05/2025

Join us today, May 27, for @trienetoscience.bsky.social's PhD thesis defense. Natanya is a graduate student in @calin.bsky.social's lab, where her research has focused on innovations in programmable nucleic acid libraries and CRISPR enrichment for molecular biology applications. 🧬

063

Reposted by Calin Plesa

The Knight Campus at the University of Oregon @uoknightcampus.bsky.social · 22/05/2025

Join us tomorrow, May 23, for Andrew Holston’s PhD thesis defense. Andrew is a graduate student in @calin.bsky.social’s lab, where his research has focused on the large-scale engineering of chimeric histidine kinases. 🧪

011

Reposted by Calin Plesa

Pascal Notin @pascalnotin.bsky.social · 08/05/2025

Have we hit a "scaling wall" for protein language models? 🤔 Our latest ProteinGym v1.3 release suggests that for zero-shot fitness prediction, simply making pLMs bigger isn't better beyond 1-4B parameters. The winning strategy? Combining MSAs & structure in multimodal models!

1247

Reposted by Calin Plesa

Max Fürst @maxfus.bsky.social · 07/05/2025

Preprint from the lab🚨 Have you ever engineered proteins to be more stable and were unhappy about your predictor's success rate? We got you covered with BoostMut! Great work led by @kerlenkorbeld.bsky.social now online at www.biorxiv.org/content/10.1... A thread 🧵

1418

Reposted by Calin Plesa

SynPlexity @synplexity.bsky.social · 26/03/2025

We unlock breakthrough insights by delivering the best training sets in biology—at unprecedented scale. youtu.be/KuCaJTPHM2o

youtu.be

Save 90% on DNA variant libraries today

YouTube video by IndieBio

035

Reposted by Calin Plesa

Carl de Boer @carldeboer.bsky.social · 26/03/2025

Join us June 19-20, 2025 in Vancouver BC for the Cascadia Advanced Genomic Technologies meeting! Featuring Keynote speaker Calin Plesa @calin.bsky.social Abstract deadline April 30🔔, but registration is capped so don't wait! 😱 de-boer-lab.github.io/CAGT_meeting/

de-boer-lab.github.io

CAGT: Cascadia Advanced Genomic Technologies

June 19-20 2025, Vancouver, BC, Canada

155

We’d love to hear your comments and feedback! Disclosure: I’m a co-founder and equity holder in @synplexity.bsky.social, which is commercializing DropSynth to empower other researchers to perform similar studies at scale. 13/n

010

Computational: There is a huge amount of data here and we have only started scratching the surface in its analysis. We’ve already started playing around with ML models for prediction and generation. 12/n

100

Experimental: this study was only an initial pilot. We’re assembling a library 10x as large and with continuing improvements in NGS, the amount of functional data that can be collected will be incredible. 11/n

100

Zooming in: Each homolog typically has ~30 mutants. This granular view lets us see not just overall trends but also some of the individual mutations driving resistance or loss of function. 9/n

130

Next, we turned to Trimethoprim resistance. We examined how each homolog behaved across 6 concentrations, with only a few still surviving at the highest concentration tested. 8/n

100

Patterns in gain-of-function mutants pointed to specific regions critical for restoring function. In DHFR many factors contribute to fitness: catalytic activity, expression level, folding, metabolic flux, coupling to thymidylate synthase, PPIs, … 7/n

100

The silver lining: gain-of-function mutants. Many poorly complementing homologs gained function with minor changes. Including mutants up to 3 a.a. away increases the fraction of complementing homologs to 91%. 6/n

100

Only 50% of the DHFR homologs were able to complement the KO strain well. What does this mean for ML models trained on unlabeled metagenomic data? Could non-functional sequences be "poisoning" the training set? Further work is needed to quantify this effect. 5/n

100

We adapted the DHFR KO strain and expression plasmid from the @kreynoldslab.bsky.social and Kortemme labs to assay the functionality of our libraries. Homologs were assayed for their ability to complement the KO or survive exposure to 6 levels of Trimethoprim inhibitor. 4/n