17 Search Results

This work introduces a three-dimensional (3D) invariant graph-to-string transformer variational autoencoders (VAE) (Vagrant) for generating molecules with accurate density functional theory (DFT)-level properties. Vagrant learns to model the joint probability distribution of a 3D molecular structure and its properties by encoding molecular structures into a 3D-aware latent space. Directed navigation through this latent space implicitly optimizes the 3D structure of a molecule, and the latent embedding can be used to condition a generative transformer to predict the candidate structure as a one-dimensional (1D) sequence. Additionally, we introduce two novel sampling methods that exploit the latent characteristics of a VAE tomore » improve performance. Here, we show that our method outperforms comparable 3D autoregressive and diffusion methods for predicting quantum chemical property values of novel molecules in terms of both sample quality and computational efficiency.« less

Chemical engineering is being rapidly transformed by the tools of data science. On the horizon, artificial intelligence (AI) applications will impact a huge swath of our work, ranging from the discovery and design of new molecules to operations and manufacturing and many areas in between. Early adoption of data science, machine learning, and early examples of AI in chemical engineering has been rich with examples of molecular data science—the application tools for molecular discovery and property optimization at the atomic scale. Here, we summarize key advances in this nascent subfield while introducing molecular data science for a broad chemical engineeringmore » readership. We introduce the field through the concept of a molecular data science life cycle and discuss relevant aspects of five distinct phases of this process: creation of curated data sets, molecular representations, data-driven property prediction, generation of new molecules, and feasibility and synthesizability considerations.« less

An implementation of attention within the variational autoencoder framework for continuous representation of molecules. The addition of attention significantly increases model performance for complex tasks such as exploration of novel chemistries.

One-carbon compounds such as methane and methanol are of increasing interest as sustainable substrates for biological production of fuels and industrial chemicals. The bacteria that carry out these conversions have been studied for many decades, but gaps exist in our knowledge of their metabolic pathways. One such gap is the difference between growth on methane and growth on methanol. Understanding such metabolism is important, since each has advantages and disadvantages as a feedstock for production of chemicals and fuels. The significance of our research is in the demonstration that the metabolic network is substantially altered in each case and inmore » the delineation of these changes. The resulting new insights into the core metabolism of this bacterium now provide an improved basis for future strain design.« less

We describe experiments that follow species dynamics and gene expression patterns in synthetic bacterial communities including species that compete for the single carbon substrate supplied, methane, and species unable to consume methane, which could only succeed through cooperative interactions. We demonstrate that these communities mostly select for two functional guilds, methanotrophs of the family Methylococcaceae and non-methanotrophic methylotrophs of the family Methylophilaceae, these taxonomic guilds outcompeting all other species included in the synthetic mix. The metatranscriptomics analysis uncovered that in both Methylococcaceae and Methylophilaceae, some of the most highly transcribed genes were the ones encoding methanol dehydrogenases (MDH). Remarkably, expressionmore » of alternative MDH genes (mxaFI versus xoxF), previously shown to be subjects to the rare Earth element switch, was found to depend on environmental conditions such as nitrogen source and methane and O₂ partial pressures, and also to be species-specific. Along with the xoxF genes, genes encoding divergent cytochromes were highly expressed in both Methylophilaceae and Methylococcaceae, suggesting their function in methanol metabolism, likely encoding proteins serving as electron acceptors from XoxF enzymes. The research presented tested a synthetic community model that is much simplified compared to natural communities consuming methane, but more complex than the previously utilized two-species model. The performance of this model identifies prominent species for future synthetic ecology experiments and highlights both advantages of this approach and the challenges that it presents.« less

Many methylotrophs, microorganisms that consume carbon compounds lacking carbon–carbon bonds, use two different systems to oxidize methanol for energy production and biomass accumulation. The MxaFI methanol dehydrogenase (MDH) contains calcium in its active site, while the XoxF enzyme contains a lanthanide in its active site. The genes encoding the MDH enzymes are differentially regulated by the presence of lanthanides. In this study, we found that the histidine kinase MxaY controls the lanthanide-mediated switch in Methylomicrobium buryatense 5GB1C. MxaY controls the transcription of genes encoding MxaFI and XoxF at least partially by controlling the transcript levels of the orphan response regulatormore » MxaB. We identify a constitutively active version of MxaY, and identify the mutated residue that may be involved in lanthanide sensing. Finally, we find evidence to suggest that tight control of active MDH production is required for wild-type growth rates.« less