Biology subjects

Kuznetsov, G.

Publications and source records attributed to Kuznetsov, G..

3 recordsLinked to original sources

bioRxiv2018

Toward machine-guided design of proteins

Proteins--molecular machines that underpin all biological life--are of significant therapeutic and industrial value. Directed evolution is a high-throughput experimental approach for improving protein function, but has difficulty escaping local maxima in the fitness landscape. Here, we investigate how supervised learning in a closed loop with DNA synthesis and high-throughput screening can be used to improve protein design. Using the green fluorescent protein (GFP) as an illustrative example, we demonstrate the opportunities and challenges of generating training datasets conducive to selecting strongly generalizing models. With prospectively designed wet lab experiments, we then validate that these models can generalize to unseen regions of the fitness landscape, even when constrained to explore combinations of non-trivial mutations. Taken together, this suggests a hybrid optimization strategy for protein design in which a predictive model is used to explore difficult-to-access but promising regions of the fitness landscape that directed evolution can then exploit at scale.

synthetic biology↗

bioRxiv2016

Millstone: Software for Multiplex Microbial Genome Analysis and Engineering

Inexpensive DNA sequencing and advances in genome editing have made computational analysis a major rate-limiting step in adaptive laboratory evolution and microbial genome engineering. We describe Millstone, a web-based platform which automates genotype comparison and visualization for projects with up to hundreds of genomic samples. To enable iterative genome engineering, Millstone allows users to design oligonucleotide libraries and create successive versions of reference genomes. Millstone is open source and easily deployable to a cloud platform, local cluster, or desktop, making it a scalable solution for any lab.

synthetic biology↗

bioRxiv2016

Optimizing complex phenotypes through model-guided multiplex genome engineering

Optimization of complex phenotypes in engineered microbial strains has traditionally been accomplished by laboratory evolution. However, only a subset of the resulting mutations may affect the phenotype of interest and many others may have unintended effects. Multiplexed genome editing can complement evolutionary approaches by creating diverse combinations of targeted changes, but in both cases it remains challenging to identify which alleles influence the desired phenotype. We present a method for identifying a minimal set of genomic modifications that optimizes a complex phenotype by combining iterative cycles of multiplex genome engineering and predictive modeling. We applied our method to the 63-codon E. coli strain C321.{Delta}A, which has 676 mutations relative to its wild-type ancestor, and identified six single nucleotide mutations that together recover 59% of the fitness defect exhibited by the strain. The resulting optimized strain, C321.DA.opt, is an improved chassis for production of proteins containing non-standard amino acids. Our data reveal how multiple cycles of multiplex automated genome engineering (MAGE) and inexpensive sequencing can generate rich genotypic and phenotypic diversity that can be combined with linear regression techniques to quantify individual allelic effects. While laboratory evolution relies on enrichment as a proxy for allelic effect, our model-guided approach is less susceptible than enrichment to bias from population dynamics and recombination efficiency. We also show that the method can identify beneficial de novo mutations that arise adventitiously. Beyond improving the fitness of C321, {Delta}A, our work provides a proof-of-principle for high-throughput quantification of individual allelic effects which can be used with any method for generating targeted genotypic diversity.

synthetic biology↗