Machine Learning Designs Proteins: Index #6

The Church lab is using machine learning to guide protein engineering efforts. In a new preprint, they claim that “as few as 24 functionally assayed mutant sequences” can be used “to build an accurate virtual fitness landscape”.

This is from Codon, my weekly newsletter. Subscribe for free.

This newsletter has changed. Research items will now be curated, non-exhaustive, and explained in a few sentences, rather than provided as a list. I will still cover peer-reviewed research, preprints, and reviews. News items, and a special bonus section, have been moved to the end of the newsletter. While I hope that this narrative format is more appealing to readers, I welcome your feedback via direct Twitter message.

🧬Featured Research

Biosynthesis of Plant-Based Medicines in Baker’s Yeast (Open Access)

In a tour de force of metabolic engineering, Prashanth Srinivasan and Christina Smolke at Stanford have engineered baker’s yeast to produce tropane alkaloids. These compounds, which are normally extracted from nightshade to treat neuromuscular disorders, were biosynthesized from starting sugars and amino acids. Constructing the synthetic pathway required more than twenty distinct proteins—taken from yeast, bacteria, plants and animals—targeted to six sub-cellular compartments, to produce the medicines. The study was published in Nature and has been widely covered in the media, including as a Nature News & Views story and in a Stanford press release.

Assembling 35(!) Pieces of DNA in a Single Tube (Open Access)

In 2018, I read a stunning paper by a team of scientists from New England Biolabs and Ginkgo Bioworks, describing an improvement to Golden Gate DNA assembly. In that paper, they described the efficient, accurate assembly of 24 DNA fragments in a single tube. Now, many of those same authors have built upon their prior study, reporting the routine assembly of 35 DNA fragments in a single reaction. To optimize the system, the authors used DNA sequencing to explore possible combinations of restriction enzymes, and then “incorporated these findings into a suite of webtools” to design more efficient DNA assembly reactions. This study was published in PLoS One.

Minicells, with Miniature Genomes, Can Build Themselves (Open Access)

Can a cell be built from scratch? A lot of people think so, and they may not be far off. A new study in Nature Communications reports that liposomes—basically sacs of phospholipids that enclose water—can be packed and programmed with a genetic blueprint. Liposomes were packaged with a loop of DNA encoding seven different genes, together comprising a biosynthetic pathway to build phospholipids. By using “fluorescence-based probes”, the researchers were able to “directly visualize membrane incorporation of synthesized phospholipids at the single vesicle level”—in other words, miniature cells, with miniature genetic programs, can help build themselves.

Fine-Tuning Gene Expression in Plants (Open Access)

Promoters are short sequences of DNA upstream of a gene that play a major role in determining how much of a protein is produced. But the rules underpinning promoters, and a gene’s expression level, have proven difficult to unravel. This week, the Patron lab reported an experimental system to rigorously investigate how different traits in a promoter—its sequence and positioning of regulatory elements, for example—impact the expression level of a gene. With these findings, a suite of “minimal” promoters were developed that can be used to more precisely tune a gene’s output. The study was published in Nucleic Acids Research. Read the press release from the Earlham Institute.

Software Package Simplifies Inter-Lab Measurements (Open Access)

The International Genetically Engineered Machine competition, or iGEM, has brought together young synthetic biology students for nearly two decades. A cornerstone of that competition is the measurement lab, whereby students carefully measure engineered cells and report their results to study how those measurements differ across laboratories and devices. A new study, in ACS Synthetic Biology, builds upon the dire need for standardization in synthetic biology, offering a software tool (written for R) that can calibrate fluorescent and plate reader measurements.

🧫 Rapid-Fire Highlights

More research & reviews worth your time

📰 #SynBio in the News