SEARCH · Search NASA

Results for “Manifold sampling”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9

All sources NASA NTRS DOE OSTI

DOE OSTI2019

Pneumatic System for the Neutron Generator Facility at SRNL

A neutron generator system equipped with a pneumatic sample transfer system, enables scientists to perform neutron activation analysis with increased efficiency and accuracy. The samples are placed in sealed containers called 'rabbits' and are passed through the system. Once sealed in the rabbit, the samples are loaded into the tubes via a hopper mechanism. The sample is launched into the generator with compressed air and captures neutrons, causing the sample to become radioactive. Next, the sample is sent to a detector, where the samples are counted for a set length of time on a Ge detector. Once the detection is completed, the sample is then sent to the trash. The system utilizes an interface to select which generator valve and detector location to send the sample too. This project is supports the future development of the neutron generator in the F wing of 773-A. A program is used to automate the passing of the rabbit through the system. The user is able to select the generator valve and detector to be utilized in the testing of the sample. Using the 'Add Rabbit' button the user can add as many sample rabbits to rabbit editor tab. Characteristics are then entered into the fields for each rabbit. The Pre-Det Cooldown and Post-Det Cooldown are intervals used to space out the time before the rabbit is sent into and out of the detector. After all characteristics are entered, the 'Submit to Hopper' button is selected thus preparing the program to run the rabbits in the inserted order. Small setup of the system using the available parts. Simulates the movement of the rabbit from the generator to the detector. Small valves are located before each manifold. To transport the rabbit from one side to the other, the valve of the sending side must be open and the receiving side must be closed. The manifold and valve in direct line with the compressor would extend to the hopper. Once the hopper is loaded, the valve would be open to send the rabbit to generator. Additional manifolds and other mechanical parts are necessary to complete the system design. To complete the construction of the system additional parts, such as the mechanical parts for the switching mechanism and the solenoid valves, are necessary. Further, to complete the automatization of the system, an Arduino is required to communicate with the valves and adjust the linear servos appropriately so that the rabbits are sent to the selected generator location and detector. This work is aimed to be completed in the next fiscal year.

42 ENGINEERING↗

DOE OSTI2019

A comprehensive modeling procedure for estimating statistical properties of forced ignition

In this work, a comprehensive modeling procedure for estimating the probability of ignition with application to high altitude relights of aircraft combustors is developed. In these configurations, an ignitor is used to introduce high-enthalpy discharge into a fuel-laden but stratified flow. Because of the inherent variabilities inflow conditions, kernel discharge process, and the chaotic turbulent flow, ignition cannot be described deterministically, but only as a probabilistic measure. The proposed modeling framework consists of three components. The turbulent flow is represented using the large eddy simulation (LES) framework. The ignition process is modeled using a manifold approach, where the initial kernel evolution is represented using a homogeneous reactor while the latter part of the evolution is represented as a competition between diffusion and chemical reactions using a flamelet-type mapping. A combined lookup table that can track the evolution of the kernel through these two distinct reaction stages is developed. The table lookup variables are solved within the LES framework. The resulting simulation tool is then embedded within an uncertainty quantification approach, where variations in the turbulent flow, as well as operating and kernel properties, are simulated using a Monte-Carlo-based sampling approach. Techniques to reduce computational cost are used to obtain a robust, numerically accurate, and physically representative model for engine relight. The method is validated using experimental data for ignition of methane/air mixtures. Due to the comprehensive nature of the modeling procedure, it is found that the simulation tool reproduces experimentally observed ignition probabilities over a wide range of operating conditions.

42 ENGINEERING↗

DOE OSTI2016

Half-metallic Co-based quaternary Heusler alloys for spintronics: Defect- and pressure-induced transitions and properties

Heusler compounds offer potential as spintronic devices due to their spin polarization and half-metallicity properties, where electron spin-majority (minority) manifold exhibits states (band gap) at the electronic chemical potential, yielding full spin polarization in a single manifold. Yet, Heuslers often exhibit intrinsic disorder that degrades its half-metallicity and spin polarization. Using density-functional theory, we analyze the electronic and magnetic properties of equiatomic Heusler (L2 1 ) CoMnCrAl and CoFeCrGe alloys for effects of hydrostatic pressure and intrinsic disorder (thermal antisites, binary swaps, and vacancies). Under pressure, CoMnCrAl undergoes a metallic transition, while half-metallicity in CoFeCrGe is retained for a limited range. Antisite disorder between Cr-Al pair in CoMnCrAl alloy is energetically the most favorable, and retains half-metallic character in Cr-excess regime. However, Co-deficient samples in both alloys undergo a transition from half-metallic to metallic, with a discontinuity in the saturation magnetization. For binary swaps, configurations that compete with the ground state are identified and show no loss of half-metallicity; however, the minority-spin band gap and magnetic moments vary depending on the atoms swapped. For single binary swaps, there is a significant energy cost in CoMnCrAl but with no loss of half-metallicity. Although a few configurations in CoFeCrGe energetically compete with the ground state, the minority-spin band gap and magnetic moments vary depending on the atoms swapped. Furthermore, this information should help in controlling these potential spintronic materials.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2024

Subcellular Feature-Based Classification of α and β Cells Using Soft X-ray Tomography

The dysfunction of α and β cells in pancreatic islets can lead to diabetes. Many questions remain on the subcellular organization of islet cells during the progression of disease. Existing three-dimensional cellular mapping approaches face challenges such as time-intensive sample sectioning and subjective cellular identification. To address these challenges, we have developed a subcellular feature-based classification approach, which allows us to identify α and β cells and quantify their subcellular structural characteristics using soft X-ray tomography (SXT). We observed significant differences in whole-cell morphological and organelle statistics between the two cell types. Additionally, we characterize subtle biophysical differences between individual insulin and glucagon vesicles by analyzing vesicle size and molecular density distributions, which were not previously possible using other methods. These sub-vesicular parameters enable us to predict cell types systematically using supervised machine learning. We also visualize distinct vesicle and cell subtypes using Uniform Manifold Approximation and Projection (UMAP) embeddings, which provides us with an innovative approach to explore structural heterogeneity in islet cells. This methodology presents an innovative approach for tracking biologically meaningful heterogeneity in cells that can be applied to any cellular system.

3D cell mapping↗

DOE OSTI2026

Prospects for Astrobiology and Technosignature Searches with the Vera C. Rubin Observatory Legacy Survey of Space and Time

The Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) will map sources in multiband colour--variability space. We present a prototype coherence-based framework for astrobiology and technosignature searches, in which candidates are treated as structured departures from natural astrophysical manifolds rather than isolated photometric outliers. We illustrate the framework with three simulated cases: five Kuiper Belt Object (KBO) surface/activity states, a grid of 649 synthetic exoplanet spectra with vegetation-red-edge-like (VRE) perturbations, and 500 synthetic multiband light curves, each projected into LSST-like observable space and analysed through colour geometry, chromatic variability, and cross-band coherence. Key results include a full-colour Mahalanobis distance $D\approx5.1$ for the weak-coma KBO state (${\sim}5σ$ in the five-dimensional colour vector), an indicative VRE coherence threshold at $f_{\rm crit}\approx0.13$, and an idealised stacking forecast reaching $5σ$ under optimistic assumptions. We show, using a small Gaia~DR3 stellar sample, that stellar colour and photometric stability may inform the prioritisation of Galactic regions for applying such coherence diagnostics.

Kovačević, Andjelka B. [Belgrade U.] (ORCID:000000↗

DOE OSTI2019

A high-speed, high-performance, microfabricated comprehensive two-dimensional gas chromatograph

A small, consumable-free, low-power, ultra-high-speed comprehensive GC×GC system consisting of microfabricated columns, nanoelectromechanical system (NEMS) cantilever resonators for detection, and a valve-based stop-flow modulator is demonstrated. The separation of a highly polar 29-component mixture covering a boiling point range of 46 to 253 °C on a pair of microfabricated columns using a Staiger valve manifold in less than 7 seconds, and just over 4 seconds after the ensemble holdup time is demonstrated with a downstream FID. The analysis time of the second dimension was 160 ms, and peak widths in the second dimension range from 10–60 ms. A peak capacity of just over 300 was calculated for a separation of just over 6 s. Data from a continuous operation testing over 40 days and 20 000 runs of the GC×GC columns with the NEMS resonators using a 4-component test set is presented. As a result, the GC×GC-NEMS resonator system generated second-dimension peak widths as narrow as 8 ms with no discernable peak distortion due to under-sampling from the detector.

47 OTHER INSTRUMENTATION↗

DOE OSTI2023

Projection pursuit adaptation on polynomial chaos expansions

Here, the present work addresses the issue of accurate stochastic approximations in high-dimensional parametric space using tools from uncertainty quantification (UQ). The basis adaptation method and its accelerated algorithm in polynomial chaos expansions (PCE) were recently proposed to construct low-dimensional approximations adapted to specific quantities of interest (QoI). The present paper addresses one difficulty with these adaptations, namely their reliance on quadrature point sampling, which limits the reusability of potentially expensive samples. Projection pursuit (PP) is a statistical tool to find the “interesting” projections in high-dimensional data and thus bypass the curse-of-dimensionality. In the present work, we combine the fundamental ideas of basis adaptation and projection pursuit regression (PPR) to propose a novel method to simultaneously learn the optimal low-dimensional spaces and PCE representation from given data. While this projection pursuit adaptation (PPA) can be entirely data-driven, the constructed approximation exhibits mean-square convergence to the solution of an underlying governing equation and thus captures the supports and probability distributions associated with the physics constraints. The proposed approach is demonstrated on a borehole problem and a structural dynamics problem, demonstrating the versatility of the method and its ability to discover low-dimensional manifolds with high accuracy with limited data. In addition, the method can learn surrogate models for different quantities of interest while reusing the same data set.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2015

Extensive sensitivity study of PWR fuel assay data - 14301

Accurate calculations of the isotopic composition of spent nuclear fuel play a fundamental role in several parts of nuclear energy. An important example is the use of burnup credit in criticality safety analysis, where the results gained from depletion calculations are used to determine the fuel composition in subsequent criticality calculations. The work presented here is a sensitivity study for depletion calculations of high-burnup fuel samples from the Vandellos II PWR that entails an unusually detailed evaluation of assay data. In addition to investigating the impacts of experimental uncertainties on the calculated isotopic composition, the consequences of various modelling approaches are studied to give quantitative estimations of the impact on accuracy. Furthermore, the reactivity of the calculated fuel composition is tracked, because this is the integral characteristic of the irradiated fuel that one is actually interested in for burnup credit applications. While the sensitivity study is performed only for one of the nine experimental samples of the Vandellos II assay data, the lessons learned lead to evaluation models with different levels of model detail for all fuel samples. The results show that the discrepancies between the measured and calculated isotopic compositions are considerably larger than the uncertainties inherent in the measurement and the calculation method would indicate. This is not unexpected, since large discrepancies are a common result of assay data recalculations. However, the unusually extensive sensitivity study performed here allows to exclude some possible sources of error: since the influence of experimental uncertainties like fuel assembly geometry and reactor parameters on the isotopic composition is shown to be small, one valuable conclusion of this study is to focus more on the measurement uncertainties of the chemical assay itself, and the nuclear data that the calculation is based upon. The study also demonstrates that using the reactivity of the depleted fuel as an additional comparison point between measurement and calculation can provide some orientation regarding the importance of the manifold influences on calculation results. Tracking this integral characteristic of the depleted fuel is very helpful in focusing on the actual impact of the depletion calculation uncertainties on criticality safety. (author)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

DOE OSTI2025

Data-Driven Compositional Optimization in Misspecified Regimes

With a manifold growth in the scale and intricacy of systems, the challenges of parametric misspecification become pronounced. These concerns are further exacerbated in compositional settings, which emerge in problems complicated by modeling risk and robustness. In “Data-Driven Compositional Optimization in Misspecified Regimes,” the authors consider the resolution of compositional stochastic optimization problems, plagued by parametric misspecification. In considering settings where such misspecification may be resolved via a parallel learning process, the authors develop schemes that can contend with diverse forms of risk, dynamics, and nonconvexity. They provide asymptotic and rate guarantees for unaccelerated and accelerated schemes for convex, strongly convex, and nonconvex problems in a two-level regime with extensions to the multilevel setting. Surprisingly, the nonasymptotic rate guarantees show no degradation from the rate statements obtained in a correctly specified regime and the schemes achieve optimal (or near-optimal) sample complexities for general T-level strongly convex and nonconvex compositional problems.

Business & Economics↗

DOE OSTI2019

Resolving Local Configurational Contributions to X-ray and Neutron Radial Distribution Functions Within Solutions of Concentrated Electrolytes – A Case Study of Concentrated NaOH

Extreme conditions of complex materials often lead to a manifold of local environments that challenge characterization and require new advances at the intersection of modern experimental and theoretical techniques. In this contribution, highly caustic and viscous aqueous NaOD solutions were characterized with a combination of X-ray and neutron radial distribution function (RDF) analyses, molecular dynamics simulations and sub-ensemble analysis. The interpretation of the experimental RDFs extends beyond the traditional understanding of atom-pair interactions and to the individual contributions of specific molecular-scale configurations (below 3.5 Å). This integrated approach yields unique insight into the complex and significant breath of ion solvation and pairing structural motifs present, the experimental sensitivity of RDFs to changes in local geometries, and the challenge of experimentally differentiating the ensemble of all superimposed local environments — a feature of increasing importance within the extreme condition of high ionic strength. This study was facilitated by the development of a chemically resistant sample cell for in-situ neutron total scattering measurements of highly caustic materials at elevated temperature and pressure conditions. Advancing these integrated capabilities is critical for ongoing model validation efforts and for extending the predictive capabilities needed for the extremes of chemical composition, aiding a variety of industrial processes that include advanced nuclear waste remediation and metal refining strategies.

Semrouni, David↗

DOE OSTI2018

Addressing Research and Development Gaps for Plasma-Material Interactions with Linear Plasma Devices

Plasma-material interactions in future fusion reactors have been identified as a knowledge gap to be dealt with before any next step device past ITER can be built. The challenges are manifold. They are related to power dissipation so that the heat fluxes to the plasma-facing components can be kept at technologically feasible levels; maximization of the lifetime of divertor plasma-facing components that allow for steadystate operation in a reactor to reach the neutron fluence required; the tritium inventory (storage) in the plasma-facing components, which can lead to potential safety concerns and reduction in the fuel efficiency; and it is related to the technology of the plasma-facing components itself, which should demonstrate structural integrity under the high temperatures and high neutron fluence. While the dissipation of power exhaust can and should be addressed in high power toroidal devices, the interaction of the plasma with the materials can be best addressed in dedicated linear devices due to their cost effectiveness and ability to address urgent research and development gaps more timely. However, new linear plasma devices are needed to investigate the PMI under fusion reactor conditions and test novel plasma-facing components. Existing linear devices are limited either in their flux, their reactor-relevant plasma transport regimes in front of the target, their fluence, or their ability to test material samples a priori exposed to high neutron fluence. The proposed Material Plasma Exposure eXperiment (MPEX) is meant to address those deficiencies and will be designed to fulfill the fusion reactor-relevant plasma parameters as well as the ability to expose a priori neutron activated materials to plasmas.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2015

MORFOMETRYKA—A NEW WAY OF ESTABLISHING MORPHOLOGICAL CLASSIFICATION OF GALAXIES

We present an extended morphometric system to automatically classify galaxies from astronomical images. The new system includes the original and modified versions of the CASGM coefficients (Concentration C{sub 1}, Asymmetry A{sub 3}, and Smoothness S{sub 3}), and the new parameters entropy, H, and spirality σ{sub ψ}. The new parameters A{sub 3}, S{sub 3}, and H are better to discriminate galaxy classes than A{sub 1}, S{sub 1}, and G, respectively. The new parameter σ{sub ψ} captures the amount of non-radial pattern on the image and is almost linearly dependent on T-type. Using a sample of spiral and elliptical galaxies from the Galaxy Zoo project as a training set, we employed the Linear Discriminant Analysis (LDA) technique to classify EFIGI (Baillard et al. 4458 galaxies), Nair and Abraham (14,123 galaxies), and SDSS Legacy (779,235 galaxies) samples. The cross-validation test shows that we can achieve an accuracy of more than 90% with our classification scheme. Therefore, we are able to define a plane in the morphometric parameter space that separates the elliptical and spiral classes with a mismatch between classes smaller than 10%. We use the distance to this plane as a morphometric index (M{sub i}) and we show that it follows the human based T-type index very closely. We calculate morphometric index M{sub i} for ∼780k galaxies from SDSS Legacy Survey–DR7. We discuss how M{sub i} correlates with stellar population parameters obtained using the spectra available from SDSS–DR7.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2016

Ru nucleation and thin film smoothness improvement with ammonia during chemical vapor deposition

This study reports the use of ammonia to inhibit the growth of previously nucleated ruthenium islands and force the nucleation of additional islands such that thinner films form as the islands coalesce with continued growth. Ruthenium films are grown at 448 K in a chemical vapor deposition process on SiO{sub 2}/Si(001) using triruthenium dodecacarbonyl, Ru{sub 3}(CO){sub 12}, with and without a constant partial pressure of ammonia. Film growth was performed at a Ru{sub 3}(CO){sub 12}/Ar pressure of 47.2 mTorr. The ammonia partial pressure varied from 0 to 27.8 mTorr. X-ray photoelectron spectroscopy was used to analyze the samples in situ. Ex situ characterization included scanning electron microscopy, atomic force microscopy, and x-ray diffraction and x-ray reflectivity. Nucleation studies limited to the first 10 min of growth revealed the maximum nanoparticle (island) density of 8.1 × 10{sup 11 }cm{sup −2} occurred at an intermediate ammonia pressure (5.25 mTorr) compared to a density of 3.1 × 10{sup 11 }cm{sup −2} for no ammonia addition. Extending film growth to 120 min and varying the ammonia partial pressure during the first 10 min followed by 5.25 mTorr ammonia pressure for the final 110 min reveals the importance of nucleation on film smoothness. A model describing the inhibition effects of ammonia during nucleation and growth is presented.

36 MATERIALS SCIENCE↗

NASA NTRS2016

Biomolecular Analysis Capability for Cellular and Omics Research on the International Space Station

International Space Station (ISS) assembly complete ushered a new era focused on utilization of this state-of-the-art orbiting laboratory to advance science and technology research in a wide array of disciplines, with benefits to Earth and space exploration. ISS enabling capability for research in cellular and molecular biology includes equipment for in situ, on-orbit analysis of biomolecules. Applications of this growing capability range from biomedicine and biotechnology to the emerging field of Omics. For example, Biomolecule Sequencer is a space-based miniature DNA sequencer that provides nucleotide sequence data for entire samples, which may be used for purposes such as microorganism identification and astrobiology. It complements the use of WetLab-2 SmartCycler"TradeMark", which extracts RNA and provides real-time quantitative gene expression data analysis from biospecimens sampled or cultured onboard the ISS, for downlink to ground investigators, with applications ranging from clinical tissue evaluation to multigenerational assessment of organismal alterations. And the Genes in Space-1 investigation, aimed at examining epigenetic changes, employs polymerase chain reaction to detect immune system alterations. In addition, an increasing assortment of tools to visualize the subcellular distribution of tagged macromolecules is becoming available onboard the ISS. For instance, the NASA LMM (Light Microscopy Module) is a flexible light microscopy imaging facility that enables imaging of physical and biological microscopic phenomena in microgravity. Another light microscopy system modified for use in space to image life sciences payloads is initially used by the Heart Cells investigation ("Effects of Microgravity on Stem Cell-Derived Cardiomyocytes for Human Cardiovascular Disease Modeling and Drug Discovery"). Also, the JAXA Microscope system can perform remotely controllable light, phase-contrast, and fluorescent observations. And upcoming confocal microscopy capability will allow for optical sectioning of biological tissues to determine microanatomical localization of biomarkers. Furthermore, NASA's geneLAB effort addresses integration of genomic, epigenomic, transcriptomic, proteomic and metabolomic datasets, by applying an innovative open source science platform for multi-investigator high throughput utilization of the ISS. In sum, the expanding ISS capability for analysis of biomolecules is enabling innovative research in a broad spectrum of areas such as cellular and molecular biology, biotechnology, tissue engineering, biomedicine, and Omics, providing manifold benefits for humanity.

Guinart-Ramirez, Y.↗

DOE OSTI2015

A catalog of visual-like morphologies in the 5 CANDELS fields using deep learning

We present a catalog of visual-like H-band morphologies of ∼50.000 galaxies (H{sub f160w} < 24.5) in the 5 CANDELS fields (GOODS-N, GOODS-S, UDS, EGS, and COSMOS). Morphologies are estimated using Convolutional Neural Networks (ConvNets). The median redshift of the sample is 〈z〉∼1.25. The algorithm is trained on GOODS-S, for which visual classifications are publicly available, and then applied to the other 4 fields. Following the CANDELS main morphology classification scheme, our model retrieves for each galaxy the probabilities of having a spheroid or a disk, presenting an irregularity, being compact or a point source, and being unclassifiable. ConvNets are able to predict the fractions of votes given to a galaxy image with zero bias and ∼10% scatter. The fraction of mis-classifications is less than 1%. Our classification scheme represents a major improvement with respect to Concentration-Asymmetry-Smoothness-based methods, which hit a 20%–30% contamination limit at high z. The catalog is released with the present paper via the Rainbow database (http://rainbowx.fis.ucm.es/Rainbow-navigator-public/).

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1994

The galactic globular cluster system

We explore correlations between various properties of Galactic globular clusters, using a database on 143 objects. Our goal is identify correlations and trends which can be used to test and constrain theoretical models of cluster formation and evolution. We use a set of 13 cluster parameters, 9 of which are independently measured. Several arguments suggest that the number of clusters still missing in the obscured regions of the Galaxy is of the order of 10, and thus the selection effects are probably not severe for our sample. Known clusters follow a power-law density distribution with a slope approximately -3.5 to -4, and an apparent core with a core radius approximately 1 kpc. Clusters show a large dynamical range in many of their properties, more so for the core parameters (which are presumably more affected by dynamical evolution) than for the half-light parameters. There are no good correlations with luminosity, although more luminous clusters tend to be more concentrated. When data are binned in luminosity, several trends emerge: more luminous clusters tend to have smaller and denser cores. We interpret this as a differential survival effect, with more massive clusters surviving longer and reaching more evolved dynamical states. Cluster core parameters and concentrations also correlate with the position in the Galaxy, with clusters closer to the Galactic center or plane being more concentrated and having smaller and denser cores. These trends are more pronounced for the fainter (less massive) clusters. This is in agreement with a picture where tidal shocks form disk or bulge passages accelerate dynamical evolution of clusters. Cluster metallicities do not correlate with any other parameter, including luminosity and velocity dispersion; the only detectable trend is with the position in the Galaxy, probably reflecting Zinn's disk-halo dichotomy. This suggests that globular clusters were not self-enriched systems. Velocity dispersions show excellent correlations with luminosity and surface brightness. Their origin is not well understood, but they may well reflect initial conditions of cluster formation, and perhaps even be used to probe the initial density perturbation spectrum on a approximately 10(exp 6) solar mass scale. Core radii and concentrations play a role of a 'second parameter' in these correlations. While a global manifold of cluster properties has a high statistical dimensionality (D greater than 4), a subset of structural, photometric, and dynamical parameters forms a statistically three-dimensional family, as expected from objects following King models; we propose to call this set of quantities the King Manifold. Some of the observed correlations may be usable as distance indicator relations for globular clusters.

Djorgovski, S.↗

DOE OSTI2020

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.↗

DOE OSTI2019

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

Colby, Sean M↗