SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
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.
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 engineering 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.
Although continuous symmetry theory has attracted increasing attention in modern chemistry, local symmetry remains under-investigated. As a consequence, the relationship between symmetry and chemical behavior is often obscured, limiting the practical use of fuzzy symmetry measures. Here, in this study, we introduce a novel framework for evaluating local symmetry based on electron density localization and present continuous symmetry representations for several representative molecules. Our approach not only quantitatively captures global symmetry but also reveals distinctive features of symmetry in a local chemical environment. The related concept, local chirality or chirotopicity, is also discussed. Overall, the proposed local symmetry and chirality measures provide valuable insights into molecular structure and structure–property relationships.
A variational autoencoder (VAE) has been developed to learn a continuous numerical, or latent, representation of molecular structure to expand reference libraries for small molecule identification. The VAE has been extended to include a chemical property decoder, trained as a multitask network, to shape the latent representation such that it assembles according to desired chemical properties. The approach is unique in its application to metabolomics and small molecule identification, focused on properties that are obtained from experimental measurements (m/z, CCS) paired with its training paradigm, which involves a cascade of transfer learning iterations. First, molecular representation is learned from a large dataset of structures with m/z labels. Next, in silico property values are used to continue training. Finally, the network is further refined by being trained with the experimental data. The trained network is used to predict chemical properties directly from structure and generate candidate structures with desired chemical properties. The network is extensible to other training data and molecular representations, and for use with other analytical platforms, for both chemical property and feature prediction as well as molecular structure generation.
Comprehensive and unambiguous identification of small molecules in complex samples will revolutionize our understanding of the role of metabolites in biological systems. Existing and emerging technologies have enabled measurement of chemical properties of molecules in complex mixtures and, in concert, are sensitive enough to resolve even stereoisomers. Despite these experimental advances, small molecule identification is inhibited by (i) chemical reference libraries (e.g. mass spectra, collision cross section, and other measurable property libraries) representing <1% of known molecules, limiting the number of possible identifications, and (ii) the lack of a method to generate candidate matches directly from experimental features (i.e. without a library). To this end, we developed a variational autoencoder (VAE) to learn a continuous numerical, or latent, representation of molecular structure to expand reference libraries for small molecule identification. We extended the VAE to include a chemical property decoder, trained as a multitask network, in order to shape the latent representation such that it assembles according to desired chemical properties. The approach is unique in its application to metabolomics and small molecule identification, with its focus on properties that can be obtained from experimental measurements (m/z, CCS) paired with its training paradigm, which involved a cascade of transfer learning iterations. First, molecular representation is learned from a large dataset of structures with m/z labels. Next, in silico property values are used to continue training, as experimental property data is limited. Finally, the network is further refined by being trained with the experimental data. This allows the network to learn as much as possible at each stage, enabling success with progressively smaller datasets without overfitting. Once trained, the network can be used to predict chemical properties directly from structure, as well as generate candidate structures with desired chemical properties. Our approach is orders of magnitude faster than first-principles simulation for CCS property prediction. Additionally, the ability to generate novel molecules along manifolds, defined by chemical property analogues, positions DarkChem as highly useful in a number of application areas, including metabolomics and small molecule identification, drug discovery and design, chemical forensics, and beyond.
Machine-readable chemical structure representations are foundational in all attempts to harness machine learning for the prediction of reactivities, selectivities, and chemical properties directly from molecular structure. The featurization of discrete chemical structures into a continuous vector space is a critical phase undertaken before model selection, and the development of new ways to quantitatively encode molecules is an active area of research. Here, we highlight the application and suitability of different representations, from expert-guided “engineered” descriptors to automatically “learned” features, in different prediction tasks relevant to organic and organometallic chemistry, where differing amounts of training data are available. These tasks include statistical models of stereo- and enantioselectivity, thermochemistry, and kinetics developed using experimental and quantum chemical data. The use of expert-guided molecular descriptors provides an opportunity to incorporate chemical knowledge, domain expertise, and physical constraints into statistical modeling. In applications to stereoselective organic and organometallic catalysis, where data sets may be relatively small and 3D-geometries and conformations play an important role, mechanistically informed features can be used successfully to obtain predictive statistical models that are also chemically interpretable. We provide an overview of several recent applications of this approach to obtain quantitative models for reactivity and selectivity, where topological descriptors, quantum mechanical calculations of electronic and steric properties, along with conformational ensembles, all feature as essential ingredients of the molecular representations used. Alternatively, more flexible, general-purpose molecular representations such as attributed molecular graphs can be used with machine learning approaches to learn the complex relationship between a structure and prediction target. This approach has the potential to out-perform more traditional representation methods such as “hand-crafted” molecular descriptors, particularly as data set sizes grow. One area where this is particularly relevant is in the use of large sets of quantum mechanical data to train quantitative structure–property relationships. A general approach toward curating useful data sets and training highly accurate graph neural network models is discussed in the context of organic bond dissociation enthalpies, where this strategy outperforms regression using precomputed descriptors. Finally, we describe how graph neural network predictions can be incorporated into mechanistically informed statistical models of chemical reactivity and selectivity. Once trained, this approach avoids the expensive computational overhead associated with quantum mechanical calculations, while maintaining chemical interpretability. We illustrate examples for which fast predictions of bond dissociation enthalpy and of the identities of radicals formed through cleavage of a molecule’s weakest bond are used in simple physical models of site-selectivity and reactivity.
Ab initio reconstruction methods have revolutionized the capabilities of small-angle X-ray scattering (SAXS), allowing the data-driven discovery of previously unknown molecular conformations, exploiting optimization heuristics and assumptions behind the composition of globular molecules. While these methods have been successful for the analysis of small particles, their impact on fibrillar assemblies has been more limited. The micrometre-range size of these assemblies and the complex interaction of their periodicities in their scattering profiles indicate that the discovery of fibril structures from SAXS measurements requires novel approaches beyond extending existing tools for molecular discovery. In this work, it is proposed to use SAXS measurements, together with diffraction theory, to infer the electron distribution of the average cross-section of a fiber. This cross-section is modeled as a discrete electron density with continuous support, allowing representations beyond binary distributions. Additional constraints, such as non-negativity or smoothness/connectedness, can also be added to the framework. The proposed approach is tested using simulated SAXS data from amyloid β fibril models and using measured data of Tobacco mosaic virus from SAXS experiments, recovering the geometry and density of the cross-sections in all cases. The approach is further tested by analyzing SAXS data from different amyloid β fibril assemblies, with results that are in agreement with previously proposed models from cryo-EM measurements. The limitations of the proposed method, together with an analysis of the robustness of the method and the combination with different experimental sources, are also discussed.
Machine learning inspired potentials continue to improve the ability for predicting structures of materials. However, many challenges still exist, particularly when calculating structures of disordered systems. These challenges are primarily due to the rapidly increasing dimensionality of the feature-vector space which in most machine-learning algorithms is dependent on the size of the structure. In this article, we present a feature-engineered approach that establishes a set of principles for representing potentials of physical structures (crystals, molecules, and clusters) in a feature space rather than a physically motivated space. Our goal in this work is to define guiding principles that optimize information storage of the physical parameters within the feature representations. In this manner, we focus on keeping the dimensionality of the feature space independent of the number of atoms in the structure. Our Structural Information Filtered Features (SIFF) potential represents structures by utilizing a feature vector of low-correlated descriptors, which correspondingly maximizes information within the descriptor. We present results of our SIFF potential on datasets composed of disordered (carbon and carbon–oxygen) clusters, molecules with C 7 O 2 H 2 stoichiometry in the GDB9-14B dataset, and crystal structures of the form (Al x Ga y In z ) 2 O 3 as proposed in the NOMAD Kaggle competition. Our potential's performance is at least comparable, sometimes significantly more accurate, and often more efficient than other well-known machine-learning potentials for structure prediction. However, primarily, we offer a different perspective on how researchers should consider opportunities in maximizing information storage for features.
Motivation: Signaling pathways capable of switching between two states are ubiquitous within living organisms. They provide the cells with the means to produce reversible or irreversible decisions. Switchlike behavior of biological systems is realized through biochemical reaction networks capable of having two or more distinct steady states which are dependent on initial conditions. Investigation of whether a certain signaling pathway can confer bistability involves a substantial amount of hypothesis testing. The cost of direct experimental testing can be prohibitive. Therefore, constraining the hypothesis space is highly bene?cial. One such methodology is based on Chemical Reaction Network Theory which uses computational techniques to rule out pathways that are not capable of bistability regardless of kinetic constant values and molecule concentrations. Although useful, these methods are complicated from both pureandcomputationalmathematicsperspectives.Thus,theiradoptionisverylimitedamongstbiologists. Results: We brought Chemical Reaction Network Theory approaches closer to experimental biologists by automating all the necessary steps in CRNT4SMBL. The input is based on SBML format which is the community standard for biological pathway communication. The tool parses SBML and derives C-graph representations of the biological pathway with mass action kinetics. Next steps involve an ef?cient search for potential saddle-node bifurcation points using an optimization technique. This type of bifurcation is important as it has the potential of acting as a switching point between two steady states. Finally, if any bifurcation points are present, numerical continuation analysis extends the equilibria branches for generating the diagram. Presence of an S-shaped bifurcation diagram indicates that the pathway acts as a bistable switch for the given optimization parameters. Availability: CRNT4SBML is available via the Python Package Index. The documentation can be found at https://crnt4sbml.readthedocs.io. CRNT4SBML is licensed under the Apache Software License 2.0. Contact: vladislav.petyuk@pnnl.gov
We report monoterpene photooxidation plays an important role in secondary organic aerosol (SOA) formation in the atmosphere. The low-volatility products can enhance new particle formation and particle growth, and thus have influences on climate feedbacks. Here, we present the results of α-pinene and Δ-3-carene photooxidation experiments conducted in continuous-flow mode in an environmental chamber under several reaction conditions. The roles of oxidants, addition of NO, and VOC molecular structure in influencing SOA yield are illustrated. SOA yield from α-pinene photooxidation shows a weak dependence on H 2 O 2 concentration, which is a proxy for HO 2 concentration. The high O/C ratios observed in the a-pinene photooxidation products suggest the production of highly oxygenated organic molecules (HOM). Addition of ozone to the chamber during low-NO x photooxidation experiments leads to higher SOA yield. With the addition of NO, the production of N-containing HOMs is enhanced and the SOA yield shows a modest, non-linear dependence on input NO concentration. Carene photooxidation leads to higher SOA yield than α-pinene under similar reaction conditions, which agrees with the lower volatility retrieved from evaporation kinetics experiments. These results improve the understanding of SOA formation from monoterpene photooxidation and could be applied to refine the representation of biogenic SOA formation in models.
Many imaging techniques for biological systems—like fixation of cells coupled with fluorescence microscopy—provide sharp spatial resolution in reporting locations of individuals at a single moment in time but also destroy the dynamics they intend to capture. In this study, these snapshot observations contain no information about individual trajectories, but still encode information about movement and demographic dynamics, especially when combined with a well-motivated biophysical model. The relationship between spatially evolving populations and single-moment representations of their collective locations is well-established with partial differential equations (PDEs) and their inverse problems. However, experimental data is commonly a set of locations whose number is insufficient to approximate a continuous-in-space PDE solution. Here, motivated by popular subcellular imaging data of gene expression, we embrace the stochastic nature of the data and investigate the mathematical foundations of parametrically inferring demographic rates from snapshots of particles undergoing birth, diffusion, and death in a nuclear or cellular domain. Toward inference, we rigorously derive a connection between individual particle paths and their presentation as a Poisson spatial process. Using this framework, we investigate the properties of the resulting inverse problem and study factors that affect quality of inference. One pervasive feature of this experimental regime is the presence of cell-to-cell heterogeneity. Rather than being a hindrance, we show that cell-to-cell geometric heterogeneity can increase the quality of inference on dynamics for certain parameter regimes. Altogether, the results serve as a basis for more detailed investigations of subcellular spatial patterns of RNA molecules and other stochastically evolving populations that can only be observed for single instants in their time evolution.
Exploring mesoscopic physical phenomena has always been a challenge for brute-force all-atom molecular dynamics simulations. Although recent advances in computing hardware have improved the accessible length scales, reaching mesoscopic timescales is still a significant bottleneck. Coarse-graining of all-atom models allows robust investigation of mesoscale physics with a reduced spatial and temporal resolution but preserves desired structural features of molecules, unlike continuum-based methods. Here, we present a hybrid bond-order coarse-grained forcefield (HyCG) for modeling mesoscale aggregation phenomena in liquid–liquid mixtures. The intuitive hybrid functional form of the potential offers interpretability to our model, unlike many machine learning based interatomic potentials. We parameterize the potential with the continuous action Monte Carlo Tree Search (cMCTS) algorithm, a reinforcement learning (RL) based global optimizing scheme, using training data from all-atom simulations. The resulting RL-HyCG correctly describes mesoscale critical fluctuations in binary liquid–liquid extraction systems. cMCTS, the RL algorithm, accurately captures the mean behavior of various geometrical properties of the molecule of interest, which were excluded from the training set. The developed potential model along with the RL-based training workflow could be applied to explore a variety of other mesoscale physical phenomena that are typically inaccessible to all-atom molecular dynamics simulations.
Satellite instruments have proven especially capable at monitoring the quantity of airborne particles in columns of atmosphere, globally. This chapter describes the principles of satellite measurements and retrieval algorithms, and surveys current instruments and their capabilities. We outline the issues associated with retrieval algorithms, such as surface characterization and aerosol proximity to clouds, and the challenges with interpretation of the results. The relationship between measured aerosol properties and climate-relevant aerosol properties simulated in models is outlined, as well as how measurements are used to evaluate models. Most space-based aerosol instruments are passive sensors that measure reflected sunlight at multiple wavelengths, some at multiple viewing angles. A few are active sensors that send out their own laser light and measure the returned signal. Except when clouds are present, the excess amount of light scattered back to space, beyond that expected from the surface and atmospheric gas, is attributed to aerosol. Satellite measurements are used in many ways in aerosol research. They often provide the only method for monitoring hazardous phenomena such as major wildfire and volcanic eruption plumes, especially in remote areas. Stable, long-term, near-global-scale satellite data records make it possible to identify regional and global aerosol trends. Aerosol radiative effects on climate can be quantified on a near-global scale and used to estimate the strength of aerosol–radiation and aerosol–cloud interactions as well as to evaluate climate model simulations of these interactions. Aerosol-type mapping from satellite imagery is helpful for source attribution, model validation, and to constrain particle light-absorption properties that are essential for radiative forcing calculations. The range of aerosol properties retrieved from satellite observations has grown considerably since the first global estimates of aerosol optical depth (τ a) over ocean were made in the late 1970s. Methods for retrieving particle size and light-absorption properties were explored in the 1990s using multispectral, multi-angle observations, and polarization in visible and near-infrared wavelengths. Sensitivity to particle light absorption, primarily from black or brown carbon content, improved with the inclusion of UV channels, and sensitivity to very thin aerosol layers in the upper troposphere and lower stratosphere was advanced with the use of limb-sounding instruments and active sensors. There are limitations to every measurement technique, including satellite aerosol remote sensing. For wide-swath, passive instruments, aerosol retrievals near clouds can present substantial challenges as far as 15 km away due to cloud-scattered light contaminating the signal. In nearly all cases, retrievals over bright snow and ice surfaces are precluded because surface reflectance uncertainties can overwhelm the aerosol signal. Similarly, meteorological cloud is identified and masked out where possible. Data from passive sensors also lack vertical resolution except those that view toward the limb or where multi-angle imagery is acquired over plumes from wildfires, erupting volcanoes, and wind-blown dust. Yet, passive sensors provide vastly more coverage than the active instruments that mitigate these issues. Particle microphysical information is qualitative from all remote sensing techniques, relying on proxies to infer particle composition, hygroscopicity, and the amount of light-absorbing material. Further, particles smaller than about 200 nm diameter cannot be distinguished from atmospheric gas molecules with remote sensing, which hinders studies of cloud condensation nuclei and their effects on clouds. Most satellite instruments dedicated to aerosol observations are in low-Earth, near-polar, sun-synchronous orbits, which means they cross the equator at the same local time each day. Most are set on cycles that repeat approximately every 16 days, which makes it difficult to monitor aerosol evolution locally. Geostationary satellites make it possible to observe changes occurring from minutes to hours over regions up to 8000 km in size, but lack coverage of high latitudes, and often provide more limited constraints on aerosol properties. Ground-truth data are vital for satellite aerosol-retrieval validation. The AErosol RObotic NETwork (AERONET) of sun photometers was created in 1993 and has become an established global network of over 350 instruments for validating satellite measurements. The network, as well as global networks of ground-based lidars, solar flux radiometers and other sun photometers, are widely used for evaluating global satellite retrievals and model simulations. NASA's Earth Observing System (EOS) program beginning in 1999 led to improvements in reliability, spatial resolution, and spectral resolution (and hence, to improved particle size discrimination and light absorption properties). Satellite payloads include advanced broad-swath and multi-angle imagers, along with the first space-based active sensor focused largely on long-term aerosol monitoring. Since about 2002, Europe's SENTINEL and operational meteorological satellite fleets are also providing sustained aerosol observations, with planned continuation until at least 2030. Satellite remote sensing instruments offer valuable data for evaluating aerosol representations in global climate models. They have been used to assess aerosol optical and physical properties, trends and distributions, and are applied increasingly as direct model constraints in data assimilation to create global aerosol reanalysis products. Aerosol optical depth is the most common quantity adopted for routine model evaluation, including multiwavelength data to loosely constrain particle-size distributions. These evaluations of multiple models have revealed general biases in their regional aerosol amounts and seasonal patterns of transport and removal. Although satellite measurements have near-global coverage, substantial errors can be introduced into the model observation comparison unless attention is paid to spatial and temporal collocation, cloud screening, subgrid-scale variability, and measurement uncertainties that vary with retrieval conditions.