Search NASA⌕ Search

SEARCH · Search NASA

Results for “Complex Network Data”

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.

At least 1,639 records · Page 91

AmeriFlux FLUXNET-1F US-xSP NEON Soaproot Saddle (SOAP)

This is the AmeriFlux Management Project (AMP) created FLUXNET-1F version of the carbon flux data for the site US-xSP NEON Soaproot Saddle (SOAP). This is the FLUXNET version of the carbon flux data for the site US-xSP NEON Soaproot Saddle (SOAP) produced by applying the standard ONEFlux (1F) software. Site Description - The Soaproot Saddle is a complex terrain of coarse hills, steep slopes and narrow drainages. With an elevation of 3274 - 4537’ this site encompasses 1438 acres of mixed conifer forests that are experiencing high levels of mortality due to native Pine beetles. At the core of this site stands a 171’ tall flux tower that collects physical and chemical properties of atmosphere and related process. Soaproot Saddle also hosts an array of sensor measurements along with field observations collected by highly trained NEON staff. The automated instrument measurements and some of the terrestrial observational safor this field site are colocted with NEON's aquatic site, Upper Big Creek, which is located just north of Soaproot Saddle's site boundaries.

Network), NEON (National Ecological Observatory [N↗

Structural Evaluation of Complex Subcomponents Manufactured by Large Scale Extrusion Deposition of Carbon Fiber Reinforced ABS

The large scale extrusion deposition process has recently been commercialized in the form of additive manufacturing machines such as the BAAM from Cincinnati Inc., which enables layered production of chopped fiber reinforced thermoplastic polymer systems. The application of this process to the automotive market is being pursued by Local Motors to enable rapid design to market cycles and build on-demand capability for a wide range of vehicle styles from a distributed network of micro-factories. This endeavor will require an understanding of the structural performance of components with complex geometries, where the properties are known to vary as a function of printing conditions, as well as orientation and shape. In order to rapidly build a database of structural performance, a beam specimen with outer dimensions of 914 x 305 x 64 mm3 was selected to be subjected to bending, torsion, and shear loading with full field displacement measurement by digital image correlation (DIC). A total of 25 different specimen configurations were designed including variation of print geometry, reinforcement by foam filling, and overwrapping with advanced composites. This work documents the results of the study and presents the relative performance of the structures in addition to the approach to simulation that is being initiated to serve as a framework for efficient design of structures manufactured in this way. The results are presented as a table of structural efficiency including stiffness and strength as functions of mass for each design. Additional results will also be provided in the form of surface displacements from DIC; future work will utilize this data to better optimize simulations for predicting structural performance of articles manufactured by this process.

Bedsole, Robert↗

Investigating Bacterial-Fungal Interactions using Fungal Highway Columns in Diverse Environments and Substrates

Bacterial-fungal interactions (BFIs) play an integral role in shaping microbial community composition, biogeochemical functions, spatial dynamics, and microbial dispersal. Mycelial networks created by filamentous fungi or other filamentous microorganisms (e.g., Oomycetes) act as 'fungal highways' that can be utilized by bacteria for transport throughout heterogeneous environments, greatly facilitating their mobility and granting them access to regions that may be challenging or impossible to reach on their own (e.g., due to air pockets within the soil). Several devices and experimental protocols have been created to study these fungal highways, including fungal highway columns. The fungal highway column designed by our group can be used for a variety of in situ or in vitro applications, as well as with diverse environmental and host-associated sample types. Herein, we describe the methods for performing experiments with these columns, including designing, printing, sterilizing, and preparing the devices. The options for analyzing data obtained from the use of these devices are also discussed here, and troubleshooting advice regarding potential pitfalls associated with experiments using fungal highway columns is offered. These devices can be used to gain a more comprehensive understanding of the diversity, mechanisms, and dynamics of fungal highway BFIs to provide valuable insights into the structural and functional dynamics within complex environments (e.g., soils) and across diverse habitats in which bacteria and fungi co-exist.

59 BASIC BIOLOGICAL SCIENCES↗

Intricate short-range order in GeSn alloys revealed by atomistic simulations with highly accurate and efficient machine-learning potentials

GeSn alloys hold promise for silicon-compatible integrated applications in electronics, photonics, and topological quantum devices. However, understanding their intricate structures using density functional theory (DFT) calculations is hindered by spatiotemporal constraints. To overcome this limitation, we develop highly accurate and efficient machine-learning interatomic potentials based on a neuroevolution potential approach with farthest point sampling on a comprehensive DFT data set. The application of the developed machine-learning potential in large-scale atomistic simulations bridges the spatiotemporal gap between modeling and advanced characterizations, and facilitates the discovery of structural intricacies in GeSn alloys. Through extensive statistical sampling, we identify a type of short-range order (SRO) that is distinguished by both its structural signature and electronic band gap from the SRO structure previously predicted. Modeling based on a large simulation cell reveals the coexistence of nano SRO domains with various degrees of ordering, demonstrating a complex spatial heterogeneity of SRO structure. Finally, our study not only reinforces the significance of fine-level structural information in alloys, but it also constitutes an effective framework for exploring SRO in a broad range of complex alloys based on highly accurate and effective machine-learning potentials.

36 MATERIALS SCIENCE↗

Machine learning-based ethylene and carbon monoxide estimation, real-time optimization, and multivariable feedback control of an experimental electrochemical reactor

Electrochemical reduction of CO 2 gas is a novel CO 2 utilization technique that has the potential to mitigate the global climate crisis caused by anthropogenic CO 2 emissions, and enable the large-scale storage of energy generated from renewable sources in the form of carbon-based chemicals and fuels. However, due to the complexity of the electrochemical reactions, the explicit first-principles models for CO2 reduction are not available yet, and there has been a limited effort to develop process modeling, optimization and control of CO 2 electrochemical reactors. To this end, a rotating cylinder electrode (RCE) reactor has been constructed at UCLA to understand the mass transfer and reaction kinetics effects separately on the productivity. In the RCE reactor, the applied potential strongly influences the reaction energetics and the electrode rotation speed affects the hydrodynamic boundary layer and modifies the film mass transfer coefficient, which involves convective and diffusive transport. Further, the present work aims to develop a multi-input multi-output (MIMO) control scheme for the RCE reactor that integrates techniques from artificial and recurrent neural network modeling, nonlinear optimization, and process controller design. Specifically, production rates of two products from the experimental reactor, ethylene and carbon monoxide, are controlled by manipulating two inputs, applied potential and catalyst rotation speed. Process dynamics and controllability are analyzed, a feedback control strategy is designed and the controllers are tuned accordingly. The experimental electrochemical cell is employed to gather data for process modeling and implement the multivariable control system. Finally, the experimental results are presented which demonstrate excellent closed-loop performance by the control system and regulation of the outputs at three different set-points including an economically-optimal set-point.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Physics-Informed Machine Learning Model for Ceramic Matrix Composite Creep

A physics-informed recurrent neural network (RNN) based surrogate model is developed to emulate the nonlinear, time-dependent constitutive behavior of ceramic matrix composites (CMCs) driven by matrix damage and constituent creep at the microscale. Physics-informed constraints are introduced into the surrogate model through regularization to ground the prediction in physics and improve its predictive capabilities. Training data is generated using the high-fidelity generalized method of cells (HFGMC) approach which calls appropriate creep and damage models for each of the constituents. This coupling permits simulating the nonlinear behavior of CMCs based on constituent response at the microscale along with microstructural features such as fiber and porosity volume fraction and fiber radius. The microscale repeating unit cell is loaded under creep fatigue conditions to replicate the material loading experienced in a turbine engine. Therefore, the RNN-based surrogate model is tasked with predicting, as a function of variable input stress sequence, temperature, and microstructural features, the resulting strain history response while satisfying physical constraints related to creep rate, isochoric inelastic deformation, and strain energy density. The trained surrogate model is shown to effectively match the strain history over quantified distributions of microstructural features and relevant loading regimes and temperatures. Neural network based surrogate models can offer efficient alternatives to running computationally intensive multiscale material models to simulate the nonlinear response of large structural models. Therefore, the presented work provides evidence towards the feasibility of developing, training, and running such models for CMCs with complex microstructures, nonlinear time-dependent material response, and under non-monotonic loading conditions.

ceramic matrix composites↗

Interpreting High-resolution Spectroscopy of Exoplanets using Cross-correlations and Supervised Machine Learning

We present a new method for performing atmospheric retrieval on ground-based, high-resolution data of exoplanets. Our method combines cross-correlation functions with a random forest, a supervised machine-learning technique, to overcome challenges associated with high-resolution data. A series of cross-correlation functions are concatenated to give a “CCF-sequence” for each model atmosphere, which reduces the dimensionality by a factor of ∼100. The random forest, trained on our grid of ∼65,000 models, provides a likelihood-free method of retrieval. The precomputed grid spans 31 values of both temperature and metallicity, and incorporates a realistic noise model. We apply our method to HARPS-N observations of the ultra-hot Jupiter KELT-9b and obtain a metallicity consistent with solar (logM = − 0.2 ± 0.2). Our retrieved transit chord temperature (T=6000{sub −200}{sup +0}K) is unreliable as strong ion lines lie outside of the extent of the training set, which we interpret as being indicative of missing physics in our atmospheric model. We compare our method to traditional nested sampling, as well as other machine-learning techniques, such as Bayesian neural networks. We demonstrate that the likelihood-free aspect of the random forest makes it more robust than nested sampling to different error distributions, and that the Bayesian neural network we tested is unable to reproduce complex posteriors. We also address the claim in Cobb et al. 2019 that our random forest retrieval technique can be overconfident but incorrect. We show that this is an artifact of the training set, rather than of the machine-learning method, and that the posteriors agree with those obtained using nested sampling.

79 ASTRONOMY AND ASTROPHYSICS↗

How Good is Good Enough?: Quantifying the Effects of Training Set Quality

There is a general consensus in the neural network community that noise in training data has a negative impact on model output; however, efforts to quantify the impact of varying levels have been limited, particularly for semantic segmentation tasks. This is a question of particular importance for remote sensing applications where the cost of producing a large training set can lead to reliance on publicly available data with varying degrees of noise. This work explores the effects of different degrees and types of training label noise on a pre-trained building extraction deep learner. Quantitative and qualitative evaluations of these effects can help inform decisions about trade-offs between the cost of producing training data and the quality of model outputs. We found that, relative to the base model, models trained with small amounts of noise showed little change in precision but achieved considerable increases in recall. Conversely, as noise levels increased, both precision and recall decreased. Precision and recall both lagged behind a model trained with pristine data. These exploratory results indicate the importance of quality control for training and, more broadly, that the relationship between degrees and types of training data noise and model performance is more complex than trade-offs between precision and recall.

Swan, Benjamin↗

Roadmap Document for Pacific Northwest National Laboratory Contribution to the Open Modeling Framework

The Cooperative Research Network (CRN) of the National Rural Electric Cooperative Association (NRECA) has identified GridLAB-D as a tool that would provide significant benefit to its member utilities. However, they have also noted that the complexity of the tool would be a significant barrier for adoption. As can often happen in complex simulation environments, as the available capabilities and flexibility increases, the usability of the software decreases except for a few “power” users; this is not unique to GridLAB-D. While GridLAB-D has expanded to a considerable user base, with a few notable exceptions (e.g., American Electric Power) most users are focused on research and development. As a result, NRECA/CRN has proposed an Open Modeling Framework (OMF) designed to make the capabilities of GridLAB-D, and other advanced grid tools, available via a web interface. This will allow utility users to access many of the capabilities of GridLAB-D, with little to no knowledge of the tool itself. Other components will be layered over the simulation engines to provide the user with business support functions, allowing full business case scenarios to be created from the technical data generated within the simulations. Because of the open availability and potential national benefit of the OMF, PNNL has been tasked with supporting NRECA/CRN’s development of the tool, with a focus on incorporating GridLAB-D within the OMF structure and expanding GridLAB-D capabilities to support OMF functions. The GridLAB-D enhancements will be provided first to the OMF developers, but will also be delivered to the wider GridLAB-D community after validation via the community repository. This report is intended to provide a roadmap for the intended enhancements to be delivered by PNNL. Seven tasks were identified in cooperation with NRECA/CRN – each is briefly discussed, including potential outcomes and deadlines.

distribution↗

MPEX AI Digital Twins

All magnetically confined plasma fusion power plant concepts (Tokamak, Spherical Tokamak, Stellarator, Mirror, ...) must exhaust the heat and plasma from the core confinement region to the material walls. The primary channel for this exhaust is through a plasma divertor which directs plasma along open magnetic field lines to a material target. The Material Plasma Exposure eXperiment (MPEX) illustrated in Figure 1, is a high-power, steady-state linear plasma device designed to produce the plasma material interaction (PMI) conditions of the divertor of future magnetic confinement fusion power plants: energy flux 20MW/m 2 , ion fluence 1031/m 2 , pulse duration 106 sec. These goals of plasma exposure in MPEX are well beyond those achieved in magnetic fusion experimental devices. Successfully achieving these high power steady state conditions for long pulses requires operational control of the heating and particle sources and the plasma flux to the walls and target. The MPEX AI Hot Spot Controller, proposed in this project, will help achieve the operational milestones of MPEX. The MPEX device will begin commissioning at the end of FY26. A smaller proto-MPEX was operated for 14,666 plasma discharges and will resume operation in September of 2025 as proto-MPEX-lite, with reduced capability, to test a new window for the Helicon plasma source. The proto-MPEX data has undergone surrogate modeling with machine learning methods (R. Archibald, 2022 IEEE International Conference on Big Data). This proto-MPEX data will be used to begin development of the AI digital twins described in this white paper. The scientific mission of MPEX is to qualify materials of different composition for use in the high energy and plasma flux conditions of a fusion power plant. The materials exposed in MPEX will in some cases be exposed to high neutron fluxes at other ORNL facilities to measure the changes to their PMI properties. The targets exposed in MPEX will be transported under vacuum to a Surface Analysis Station (SAS). The SAS will be equipped with the following diagnostics: Focused Ion Beam (FIB) for trench milling, 100-400 angstrom resolution scanning electron microscope (SEM), surface mapping x-ray spectrometer, high resolution camera, and a future upgrade to a laser induced breakdown spectroscopy quadruple mass spectrometer (LIBS-QMS). The MPEX experiments will generate diverse pre- and post-exposure measurement data of detailed material properties down to the crystal grain level in 3D for post-exposure assessment of PMI damage (e.g. cracking, melting, erosion and redeposition of the material). Physics models for the PMI, and how the material composition and manufacturing impact its performance under high energy plasma exposure, need to be validated with MPEX data to guide the selection of new candidate materials. Our vision for the MPEX AI Digital Twins project is to supply experimental and physics model simulation data to train Artificial Intelligence (AI) models for data processing, analysis, operational control, PMI and materials simulation to maximize the scientific output of the MPEX device. Ultimately, an AI digital twin of MPEX material assessment metrics for tested and synthetic material types with simulated PMI will be trained by the AI Modeling Teams on the experimental and physics simulation data submitted to the American Science Cloud by this project. A purely empirical search for the best material is inefficient given the finite number of samples that can be tested on MPEX. In order to expand the material properties database for training the MPEX Material Assessment AI Digital Twin, and to gain physics understanding of the PMI processes, physics models of the material properties and PMI processes are required. The physics simulations provide detailed simulation data, like impact angles for plasma ions, sputtering yields, transport of the ionized sputtered target material in the plasma, and redeposition locations. This simulation data expands the measurement data for deeper physics understanding. The experimental data is essential to validate the PMI and material structure simulation models. The validated models can then be used to generate new simulation data of MPEX material assessments for synthetic material compositions that have not been exposed in MPEX. These predictive simulations, plus the whole experimental dataset, will be used to train the MPEX Material Assessment AI Digital Twin allowing a rapid generative AI search for new materials with reduced PMI damage by interpolating the domain of the training set. These new optimum materials can be simulated with the physics codes and/or tested in MPEX. The ability of AI neural networks to interpolate multi-dimensional parameter spaces and generate virtual data is exploited for a more efficient search for optimum materials. The advent of the Transformational AI Models Consortium (TAIMC) is an opportunity to engage with state of the art private and public AI developers to achieve the goals of the AI digital twins and AI accelerated physics models proposed in this project. Our partners at ORNL from the Advance Scientific Computing Research (ASCR) organization will collaborate in accelerating the integrated plasma material interaction simulation framework. This simulation framework will provide a platform for generating simulation data across a range of physical fidelities, including hybrid methods that produce multi-fidelity results. This data will be leveraged for AI model development, both for generation of surrogates and the automation of simulation campaigns. A part of the research below will include collaborative efforts with the TAIMC to (i) adapt data storage approaches to ensure AI-readiness, (ii) provide a protypical exemplar to inform and exercise constructed workflows, and (iii) generate and share data, using the TAIMC unified AI data standard, for foundational models that will be trained from multiple sources across the DOE complex. We will also collaborate with the TAIMC, as well as the planned AI modeling teams, to develop approaches for reducing the cost of data generation. These include tailored multi-fidelity approaches as well as fine-tuning strategies to augment general, large-scale foundational models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Deep Learning Enabled Strain Mapping of Single-Atom Defects in Two-Dimensional Transition Metal Dichalcogenides with Sub-Picometer Precision

Two-dimensional (2D) materials offer an ideal platform to study the strain fields induced by individual atomic defects, yet challenges associated with radiation damage have so far limited electron microscopy methods to probe these atomic-scale strain fields. In this work, we demonstrate an approach to probe single-atom defects with sub-picometer precision in a monolayer 2D transition metal dichalcogenide, WSe 2–2x Te 2x . We utilize deep learning to mine large data sets of aberration-corrected scanning transmission electron microscopy images to locate and classify point defects. By combining hundreds of images of nominally identical defects, we generate high signal-to-noise class averages which allow us to measure 2D atomic spacings with up to 0.2 pm precision. Our methods reveal that Se vacancies introduce complex, oscillating strain fields in the WSe 2–2x Te 2x lattice that correspond to alternating rings of lattice expansion and contraction. These results indicate the potential impact of computer vision for the development of high-precision electron microscopy methods for beam-sensitive materials.

2D materials↗

Neural network acceleration of large-scale structure theory calculations

Here, we make use of neural networks to accelerate the calculation of power spectra required for the analysis of galaxy clustering and weak gravitational lensing data. For modern perturbation theory codes, evaluation time for a single cosmology and redshift can take on the order of two seconds. In combination with the comparable time required to compute linear predictions using a Boltzmann solver, these calculations are the bottleneck for many contemporary large-scale structure analyses. Here, we construct neural network-based surrogate models for Lagrangian perturbation theory (LPT) predictions of matter power spectra, real and redshift space galaxy power spectra, and galaxy-matter cross power spectra that attain ~ 0.1% (at one sigma) accuracy over a broad range of scales in a ωCDM parameter space. The neural network surrogates can be evaluated in approximately one millisecond, a factor of 1000 times faster than the full Boltzmann code and LPT computations. In a simulated full-shape redshift space galaxy power spectrum analysis, we demonstrate that the posteriors obtained using our surrogates are accurate compared to those obtained using the full LPT model. We make our surrogate models public at https://github.com/sfschen/EmulateLSS, so that others may take advantage of the speed gains they provide to enable rapid iteration on analysis settings, something that is essential in complex contemporary large-scale structure analyses.

79 ASTRONOMY AND ASTROPHYSICS↗

An Ultrahigh Resolution Structure of TEM-1 beta-Lactamase Suggests a Role for Glu166 as the General Base in Acylation

Although TEM-1 {beta}-lactamase is among the best studied enzymes, its acylation mechanism remains controversial. To investigate this problem, the structure of TEM-1 in complex with an acylation transition-state analogue was determined at ultrahigh resolution (0.85 {angstrom}) by X-ray crystallography. The quality of the data was such as to allow for refinement to an R-factor of 9.1% and an R{sub free} of 11.2%. In the resulting structure, the electron density features were clear enough to differentiate between single and double bonds in carboxylate groups, to identify multiple conformations that are occupied by residues and loops, and to assign 70% of the protons in the protein. Unexpectedly, even at pH 8.0 where the protein was crystallized, the active site residue Glu166 is clearly protonated. This supports the hypothesis that Glu166 is the general base in the acylation half of the reaction cycle. This structure suggests that Glu166 acts through the catalytic water to activate Ser70 for nucleophilic attack on the {beta}-lactam ring of the substrate. The hydrolytic mechanism of class A {beta}-lactamases, such as TEM-1, appears to be symmetrical, as are the serine proteases. Apart from its mechanistic implications, this atomic resolution structure affords an unusually detailed view of the structure, dynamics, and hydrogen-bonding networks of TEM-1, which may be useful for the design of inhibitors against this key antibiotic resistance target.

59 BASIC BIOLOGICAL SCIENCES↗

Probing substrate water access through the O1 channel of Photosystem II by single site mutations and membrane inlet mass spectrometry

Abstract Light-driven water oxidation by photosystem II sustains life on Earth by providing the electrons and protons for the reduction of CO 2 to carbohydrates and the molecular oxygen we breathe. The inorganic core of the oxygen evolving complex is made of the earth-abundant elements manganese, calcium and oxygen (Mn 4 CaO 5 cluster), and is situated in a binding pocket that is connected to the aqueous surrounding via water-filled channels that allow water intake and proton egress. Recent serial crystallography and infrared spectroscopy studies performed with PSII isolated fromThermosynechococcus vestitus(T. vestitus) support that one of these channels, the O1 channel, facilitates water access to the Mn 4 CaO 5 cluster during its S 2 →S 3 and S 3 →S 4 →S 0 state transitions, while a subsequent CryoEM study concluded that this channel is blocked in the cyanobacteriumSynechocystis sp.PCC 6803, questioning the role of the O1 channel in water delivery. Employing site-directed mutagenesis we modified the two O1 channel bottleneck residues D1-E329 and CP43-V410 (T. vestitusnumbering) and probed water access and substrate exchange via time resolved membrane inlet mass spectrometry. Our data demonstrates that water reaches the Mn 4 CaO 5 cluster via the O1 channel in both wildtype and mutant PSII. In addition, the detailed analysis provides functional insight into the intricate protein-water-cofactor network near the Mn 4 CaO 5 cluster that includes the pentameric, near planar ‘water wheel’ of the O1 channel.

Plant Sciences↗

Conservation of Species- and Trait-Based Modeling Network Interactions in Extremely Acidic Microbial Community Assembly

Understanding microbial interactions is essential to decipher the mechanisms of community assembly and their effects on ecosystem functioning, however, the conservation of species- and trait-based network interactions along environmental gradient remains largely unknown. Here, by using the network-based analyses with three paralleled data sets derived from 16S rRNA gene pyrosequencing, functional microarray, and predicted metagenome, we test our hypothesis that the network interactions of traits are more conserved than those of taxonomic measures, with significantly lower variation of network characteristics along the environmental gradient in acid mine drainage. The results showed that although the overall network characteristics remained similar, the structural variation was significantly lower at trait levels. The higher conserved individual node topological properties at trait level rather than at species level indicated that the responses of diverse traits remained relatively consistent even though different species played key roles under different environmental conditions. Additionally, the randomization tests revealed that it could not reject the null hypothesis that species-based correlations were random, while the tests suggested that correlation patterns of traits were non-random. Furthermore, relationships between trait-based network characteristics and environmental properties implied that trait-based networks might be more useful in reflecting the variation of ecosystem function. Taken together, our results suggest that deterministic trait-based community assembly results in greater conservation of network interaction, which may ensure ecosystem function across environmental regimes, emphasizing the potential importance of measuring the complexity and conservation of network interaction in evaluating the ecosystem stability and functioning.

59 BASIC BIOLOGICAL SCIENCES↗

Multi-omic data integration enables discovery of hidden biological regularities

Rapid growth in size and complexity of biological data sets has led to the ‘Big Data to Knowledge’ challenge. We develop advanced data integration methods for multi-level analysis of genomic, transcriptomic, ribosomal profiling, proteomic and fluxomic data. First, we show that pairwise integration of primary omics data reveals regularities that tie cellular processes together in Escherichia coli: the number of protein molecules made per mRNA transcript and the number of ribosomes required per translated protein molecule. Second, we show that genome-scale models, based on genomic and bibliomic data, enable quantitative synchronization of disparate data types. Integrating omics data with models enabled the discovery of two novel regularities: condition invariant in vivo turnover rates of enzymes and the correlation of protein structural motifs and translational pausing. These regularities can be formally represented in a computable format allowing for coherent interpretation and prediction of fitness and selection that underlies cellular physiology.

59 BASIC BIOLOGICAL SCIENCES↗

Deep learning to generate in silico chemical property libraries and candidate molecules for small molecule identification in complex samples

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.

Colby, Sean M.↗