Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Deep Direct Use”

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 271 records · Page 15
All sourcesNASA NTRSDOE OSTI
NASA NTRS2011

Thrust Stand Measurements Using Alternative Propellants in the Microwave Assisted Discharge Inductive Plasma Accelerator

Storable propellants (for example water, ammonia, and hydrazine) are attractive for deep space propulsion due to their naturally high density at ambient interplanetary conditions, which obviates the need for a cryogenic/venting system. Water in particular is attractive due to its ease of handling and availability both terrestrially and extra-terrestrially. While many storable propellants are reactive and corrosive, a propulsion scheme where the propellant is insulated from vulnerable (e.g. metallic) sections of the assembly would be well-suited to process these otherwise incompatible propellants. Pulsed inductive plasma thrusters meet this criterion because they can be operated without direct propellant-electrode interaction. During operation of these devices, electrical energy is capacitively stored and then discharged through an inductive coil creating a time-varying current in the coil that interacts with a plasma covering the face of the coil to induce a plasma current. Propellant is accelerated and expelled at a high exhaust velocity (O(10-100 km/s)) by the Lorentz body force arising from the interaction of the magnetic field and the induced plasma current. While this class of thruster mitigates the life-limiting issues associated with electrode erosion, many pulsed inductive plasma thrusters require high pulse energies to inductively ionize propellant. The Microwave Assisted Discharge Inductive Plasma Accelerator (MAD-IPA) is a pulsed inductive plasma thruster that addressees this issue by partially ionizing propellant inside a conical inductive coil before the main current pulse via an electron cyclotron resonance (ECR) discharge. The ECR plasma is produced using microwaves and a static magnetic field from a set of permanent magnets arranged to create a thin resonance region along the inner surface of the coil, restricting plasma formation, and in turn current sheet formation, to a region where the magnetic coupling between the plasma and the theta-pinch coil is high. The use of a conical theta-pinch coil also serves to provide neutral propellant containment and plasma plume focusing that is improved relative to the more common planar geometry of the Pulsed Inductive Thruster (PIT). In this paper, we describe thrust stand measurements performed to evaluate the specific impulse and thrust efficiency of the MAD-IPA for a variety of propellants. Propellants tested include both widely-used, non-reactive noble gases like argon, and rarely-used propellants such as water, hydrazine and ammonia. Dependencies of impulse data on propellant species are discussed in the context of the current sheet formation and electromagnetic plasma acceleration processes.

Hallock, Ashley K.↗
NASA NTRS1990

Observations of reconnected flux tubes within the midaltitude cusp

The paper presents three events interpreted as reconnected flux tubes that correspond to the extensions of FTEs which have penetrated deep into the magnetosphere down to the midaltitudes of the polar cusp. Low-energy plasma, high-energy plasma, magnetic fields, and electric fields are used to identify the signatures of reconnected flux tubes. Characteristics of spatial scale, time duration, and frequency of occurrence between flux transfer events and midaltitude cusp reconnected flux tubes are shown to be consistent, although they differ in the direction of motion. However, the merging cell topology and the interplanetary magnetic field effect can explain this difference. Larger-scale events can be explained by motion of the cusp resulting from a quasi-steady reconnection process. The field-aligned currents associated with reconnected flux tubes at midaltitudes within the cusp are shown to be consistent with twisting of magnetic field lines and with closure by Pedersen currents. It is considered possible that what appears to be field-aligned currents closing by Pedersen ionospheric currents may also be interpreted as currents carried by Alfven waves.

Saflekos, N. A.↗
NASA NTRS1992

Chemical environments of submarine hydrothermal systems

Perhaps because black-smoker chimneys make tremendous subjects for magazine covers, the proposal that submarine hydrothermal systems were involved in the origin of life has caused many investigators to focus on the eye-catching hydrothermal vents. In much the same way that tourists rush to watch the spectacular eruptions of Old Faithful geyser with little regard for the hydrology of the Yellowstone basin, attention is focused on the spectacular, high-temperature hydrothermal vents to the near exclusion of the enormous underlying hydrothermal systems. Nevertheless, the magnitude and complexity of geologic structures, heat flow, and hydrologic parameters which characterize the geyser basins at Yellowstone also characterize submarine hydrothermal systems. However, in the submarine systems the scale can be considerably more vast. Like Old Faithful, submarine hydrothermal vents have a spectacular quality, but they are only one fascinating aspect of enormous geologic systems operating at seafloor spreading centers throughout all of the ocean basins. A critical study of the possible role of hydrothermal processes in the origin of life should include the full spectrum of probable environments. The goals of this chapter are to synthesize diverse information about the inorganic geochemistry of submarine hydrothermal systems, assemble a description of the fundamental physical and chemical attributes of these systems, and consider the implications of high-temperature, fluid-driven processes for organic synthesis. Information about submarine hydrothermal systems comes from many directions. Measurements made directly on venting fluids provide useful, but remarkably limited, clues about processes operating at depth. The oceanic crust has been drilled to approximately 2.0 km depth providing many other pieces of information, but drilling technology has not allowed the bore holes and core samples to reach the maximum depths to which aqueous fluids circulate in oceanic crust. Such determinations rely on studies of pieces of deep oceanic crust uplifted by tectonic forces such as along the Southwest Indian Ridge, or more complete sections of oceanic crust called ophiolite sequences which are presently exposed on continents owing to tectonic emplacement. Much of what is thought to happen in submarine hydrothermal systems is inferred from studies of ophiolite sequences, and especially from the better-exposed ophiolites in Oman, Cyprus and North America. The focus of much that follows is on a few general features: pressure, temperature, oxidation states, fluid composition and mineral alteration, because these features will control whether organic synthesis can occur in hydrothermal systems.

Shock, Everett L.↗
NASA NTRS2006

Adaptive Nulling for the Terrestrial Planet Finder Interferometer

Deep, stable starlight nulls are needed for the direct detection of Earth-like planets and require careful control of the intensity and phases of the beams that are being combined. We are testing a novel compensator based on a deformable mirror to correct the intensity and phase at each wavelength and polarization across the nulling bandwidth. We have successfully demonstrated intensity and phase control using a deformable mirror across a 100nm wide band in the near-IR, and are in the process of conducting experiments in the mid-IR wavelengths. This paper covers the current results and in the mid-IR.

amplitude and phase correction↗
NASA NTRS2015

Internal NASA Study: NASAs Protoflight Research Initiative

The NASA Protoflight Research Initiative is an internal NASA study conducted within the Office of the Chief Engineer to better understand the use of Protoflight within NASA. Extensive literature reviews and interviews with key NASA members with experience in both robotic and human spaceflight missions has resulted in three main conclusions and two observations. The first conclusion is that NASA's Protoflight method is not considered to be "prescriptive." The current policies and guidance allows each Program/Project to tailor the Protoflight approach to better meet their needs, goals and objectives. Second, Risk Management plays a key role in implementation of the Protoflight approach. Any deviations from full qualification will be based on the level of acceptable risk with guidance found in NPR 8705.4. Finally, over the past decade (2004 - 2014) only 6% of NASA's Protoflight missions and 6% of NASA's Full qualification missions experienced a publicly disclosed mission failure. In other words, the data indicates that the Protoflight approach, in and of it itself, does not increase the mission risk of in-flight failure. The first observation is that it would be beneficial to document the decision making process on the implementation and use of Protoflight. The second observation is that If a Project/Program chooses to use the Protoflight approach with relevant heritage, it is extremely important that the Program/Project Manager ensures that the current project's requirements falls within the heritage design, component, instrument and/or subsystem's requirements for both the planned and operational use, and that the documentation of the relevant heritage is comprehensive, sufficient and the decision well documented. To further benefit/inform this study, a recommendation to perform a deep dive into 30 missions with accessible data on their testing/verification methodology and decision process to research the differences between Protoflight and Full Qualification missions' Design Requirements and Verification & Validation (V&V) (without any impact or special request directly to the project).

Protoflight↗
NASA NTRS2002

Knowledge Preservation for Design of Rocket Systems

An engineer at NASA Lewis RC presented a challenge to us at Southern University. Our response to that challenge, stated circa 1993, has evolved into the Knowledge Preservation Project which is here reported. The stated problem was to capture some of the knowledge of retiring NASA engineers and make it useful to younger engineers via computers. We evolved that initial challenge to this - design a system of tools such that, with this system, people might efficiently capture and make available via commonplace computers, deep knowledge of retiring NASA engineers. In the process of proving some of the concepts of this system, we would (and did) capture knowledge from some specific engineers and, so, meet the original challenge along the way to meeting the new. Some of the specific knowledge acquired, particularly that on the RL- 10 engine, was directly relevant to design of rocket engines. We considered and rejected some of the techniques popular in the days we began - specifically "expert systems" and "oral histories". We judged that these old methods had too high a cost per sentence preserved. That cost could be measured in hours of labor of a "knowledge professional". We did spend, particularly in the grant preceding this one, some time creating a couple of "concept maps", one of the latest ideas of the day, but judged this also to be costly in time of a specially trained knowledge-professional. We reasoned that the cost in specialized labor could be lowered if less time were spent being selective about sentences from the engineers and in crafting replacements for those sentences. The trade-off would seem to be that our set of sentences would be less dense in information, but we found a computer-based way around this seeming defect. Our plan, details of which we have been carrying out, was to find methods of extracting information from experts which would be capable of gaining cooperation, and interest, of senior engineers and using their time in a way they would find worthy (and, so, they would give more of their time and recruit time of other engineers as well). We studied these four ways of creating text: 1) the old way, via interviews and discussions - one of our team working with one expert, 2) a group-discussion led by one of the experts themselves and on a topic which inspires interaction of the experts, 3) a spoken dissertation by one expert practiced in giving talks, 4) expropriating, and modifying for our system, some existing reports (such as "oral histories" from the Smithsonian Institution).

Moreman, Douglas↗
NASA NTRS2011

Deep Moonquake Focal Mechanisms: Recovery and Implications

A defining characteristic of deep moonquakes is their tendency to occur with tidal periodicity, prompting previous studies to infer that they are related to the buildup and release of tidal stress within the Moon. In studies of tidal forcing, a key constraint is the focal mechanism: the fault parameters describing the type of failure moonquakes represent. The quality of the lunar seismic data and the limited source/receiver geometries of the Apollo seismic network prohibit the determination of deep moonquake fault parameters using first-motion polarities, as is typically done in terrestrial seismology. Without being able to resolve tidal stress onto a known failure plane, we can examine only gross qualities of the tidal stress tensor with respect to moonquake occurrence, so we cannot fully address the role of tidal stress in moonquake generation. We will examine the extent to which shear (S) and compression (P) wave amplitude ratios can constrain moonquake fault geometry by determining whether, for a given cluster, there exists a focal mechanism that can produce a radiation pattern consistent with the amplitudes measured by the Apollo instruments. Amplitudes are read in the ray coordinate frame, directly from seismograms for which the P and S arrivals are clearly identifiable on all long-period channels of the four Apollo stations. We apply an empirical station correction to account for site effects and the differences between P- and S-wave attenuation. Instead of focusing on the best fitting solution only, we formulate the inverse problem using a falsification criterion: all source orientations that do not reproduce the observed SV/P ratios within an error margin derived from the uncertainty of amplitude readings are rejected. All others are accepted as possible solutions. The inversion is carried out using an exhaustive grid search on a regular grid with predefined step size, encompassing all possible combinations of strike, dip and slip. To assess the sensitivity of the inversion for the uncertainty of the lunar interior structure, we carry out repeated inversions with different velocity structures. Our data set consist of a total of 106 events from 25 deep moonquake clusters. The largest contribution of 37 events originates from the most active cluster, A001, while other clusters are represented by 1 to 9 events. Since the definition of a cluster implies that all events share the same source orientation, a comparison of the inversion results of all events from one cluster will reduce ambiguities of the inversion. Once we obtain a suite of fault parameters for a given source, we can attempt to further constrain the focal mechanism with refined analyses of tidal stresses and predictions based on synthetic seismograms.

Knapmeyer, Martin↗
NASA NTRS1989

Environmental Projects. Volume 8: Modifications of wastewater evaporation ponds

The Goldstone Deep Space Communications Complex (GDSCC), located in the Mojave Desert about 45 miles north of Barstow, California, and about 160 miles northeast of Pasadena, is part of NASA's Deep Space Network, one of the world's largest and most sensitive scientific telecommunications and radio navigation networks. The Goldstone Complex is managed, technically directed, and operated for NASA by the Jet Propulsion Laboratory (JPL) of the California Institute of Technology in Pasadena, California. Activities at the GDSCC are carried out in support of seven parabolic dish antennas. These activities may give rise to environmental hazards: use of hazardous chemicals, asbestos, and underground storage tanks as well as the generation of hazardous wastes and the disposal of wastewater. Federal, state, and local laws governing the management of hazardous substances, asbestos, underground storage tanks and wastewater disposal have become so complex there is a need to devise specific programs to comply with the many regulations that implement these laws. In support of the national goal of the preservation of the environment and the protection of human health and safety, NASA, JPL, and the GDSCC have adopted a position that their operating installations shall maintain a high level of compliance with these laws. One of the environmental problems at the GDSCC involved four active, operational, wastewater evaporation ponds designed to receive and evaporate sewage effluent from upstream septic tank systems. One pair of active wastewater evaporation ponds is located at Echo Site, while another operational pair is at Mars Site.

Source record↗
NASA NTRS1991

Stratospheric soot - Cloud forming properties and transport processes

Soot enters the stratosphere either from the troposphere in deep thunderstorm convection, predominantly in the tropics, from midlatitude forest fires, and directly from aircraft flying above the tropopause. Particles emitted from jet aircraft are relatively ineffective as cloud droplet condensation nuclei (less than 1 percent) compared with those from forest fires (above 50 percent). Chemical and other modification during aging of the particles may influence these results in some circumstances; cloud nucleating properties can be used as a soot tracer. Such aerosols, as they grow to haze drops and ilute in increasing humidity, may form ice particles in the cold polar stratosphere or in tropical cumuli anvils at 18 to 20 km altitude at temperatures as low as -90 C. Ice may also form as particles are removed by descent through troposphere folds and enter midlatitude cirrus at levels below -40 C. In the stratosphere, ice particles may provide surfaces that catalyze heterogeneous chemical reactions; in the upper troposphere, the optical depths of cirrus clouds may be significantly increased.

Hallett, John↗
NASA NTRS1995

The Telecommunications and Data Acquisition Report

This quarterly publiction provides archival reports on developments in programs managed by JPL Telecommunications and Mission Operations Directorate (TMOD), which now includes the former communications and Data Acquisition (TDA) Office. In space communications, radio navigation, radio science, and ground-based radio and radar astronomy, it reports on activities of the Deep Space Network (DSN) in planning, supporting research and technology, implementation, and operations. Also included are standards activity at JPL for space data and information systems and reimbursable DSN work performed for other space agencies through NASA. The Orbital Debris Radar Program, funded by the Office of Space Systems Development, makes use of the planetary radar capability when the antennas are configured at science instruments making direct observations of planets, their satellites, and asteroids of our solar system.

Yuen, Joseph H.↗
NASA NTRS2010

Satellite Gravimetry Applied to Drought Monitoring

Near-surface wetness conditions change rapidly with the weather, which limits their usefulness as drought indicators. Deeper stores of water, including root-zone soil wetness and groundwater, portend longer-term weather trends and climate variations, thus they are well suited for quantifying droughts. However, the existing in situ networks for monitoring these variables suffer from significant discontinuities (short records and spatial undersampling), as well as the inherent human and mechanical errors associated with the soil moisture and groundwater observation. Remote sensing is a promising alternative, but standard remote sensors, which measure various wavelengths of light emitted or reflected from Earth's surface and atmosphere, can only directly detect wetness conditions within the first few centimeters of the land s surface. Such sensors include the Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) C-band passive microwave measurement system on the National Aeronautic and Space Administration's (NASA) Aqua satellite, and the combined active and passive L-band microwave system currently under development for NASA's planned Soil Moisture Active Passive (SMAP) satellite mission. These instruments are sensitive to water as deep as the top 2 cm and 5 cm of the soil column, respectively, with the specific depth depending on vegetation cover. Thermal infrared (TIR) imaging has been used to infer water stored in the full root zone, with limitations: auxiliary information including soil grain size is required, the TIR temperature versus soil water content curve becomes flat as wetness increases, and dense vegetation and cloud cover impede measurement. Numerical models of land surface hydrology are another potential solution, but the quality of output from such models is limited by errors in the input data and tradeoffs between model realism and computational efficiency. This chapter is divided into eight sections, the next of which describes the theory behind satellite gravimetry. Following that is a summary of the GRACE mission and how hydrological information is gleaned from its gravity products. The fourth section provides examples of hydrological science enabled by GRACE. The fifth and sixth sections list the challenging aspects of GRACE derived hydrology data and how they are being overcome, including the use of data assimilation. The seventh section describes recent progress in applying GRACE for drought monitoring, including the development of new soil moisture and drought indicator products, and that is followed by a discussion of future prospects in satellite gravimetry based drought monitoring.

Rodell, Matthew↗
NASA NTRS2022

Agent-Based Modeling of Microbes in Space

Space is tough on organisms. Microorganisms traveling to space experience stress from environmental features such as ionizing radiation and lack of normal gravity, and much remains unknown about the mechanisms by which those environmental features affect microbial physiology. Microbes experience changes in gravity not directly but rather through changes in their fluid environment, and deep-space particle radiation causes cell damage that is complex but rare. Computational modeling at the single-cell level (agent-based modeling) can allow us to probe the spatially heterogeneous processes that characterize space stresses, to gain insight into the relationships of microbial cells with their environments and with each other. Here we present two software packages for simulating microbial population dynamics in space conditions: CAMDLES and AMMPER. Microbes growing in liquid culture medium in the microgravity of an orbital space station experience a quiescent, poorly-mixed fluid environment. CAMDLES (CFD-DEM Artificial Microgravity Developments for Living Ecosystem Simulation) simultaneously simulates biological, chemical, and mechanical processes to predict microbial ecological dynamics in microgravity, and in the rotating culture vessels used to create an artificial microgravity environment in the lab. Initial results demonstrate that the growth of a cross-feeding microbial consortium, dependent on the exchange of soluble metabolites, is sensitive to the initial spatial distribution of cells, and grows differently in real versus artificial microgravity. Microbial populations exposed to deep-space radiation experience spatially and temporally heterogeneous damage from the traversal of high-energy particles. AMMPER (Agent-Based Model for Microbial Populations Exposed to Radiation) pairs a 3d model of energy deposition along a radiation particle track with a microbial population growth and damage model to predict the effects of localized radiation damage on population-level responses. It includes a user-friendly graphical interface. AMMPER growth curves recapitulate experimental results, and allow comparison between direct effects (DNA damage) and indirect effects (reactive oxygen species generation, metabolic impairment) of radiation.

microbiology↗
NASA NTRS1994

The Mars Observer differential one-way range demonstration

Current methods of angular spacecraft positioning using station differenced range data require an additional observation of an extragalactic radio source (quasar) to estimate the timing offset between the reference clocks at the two Deep Space Stations. The quasar observation is also used to reduce the effects of instrumental and media delays on the radio metric observable by forming a difference with the spacecraft observation (delta differential one-way range, delta DOR). An experiment has been completed using data from the Global Positioning System satellites to estimate the station clock offset, eliminating the need for the quasar observation. The requirements for direct measurement of the instrumental delays that must be made in the absence of a quasar observation are assessed. Finally, the results of the 'quasar-free' differential one-way range, or DOR, measurements of the Mars Observer spacecraft are compared with those of simultaneous conventional delta DOR measurements.

Kroger, P. M.↗
NASA NTRS2008

Microwave Radiometers from 0.6 to 22 GHz for Juno, a Polar Orbiter around Jupiter

A compact radiometer instrument is under development at JPL for Juno, the next NASA New Frontiers mission, scheduled to launch in 2011. This instrument is called the MWR (MicroWave Radiometer), and its purpose is to measure the thermal emission from Jupiter's atmosphere at selected frequencies from 0.6 to 22 GHz. The objective is to measure the distributions and abundances of water and ammonia in Jupiter's atmosphere, with the goal of understanding the previously unobserved dynamics of the subcloud atmosphere, and to discriminate among models for planetary formation in our solar system. The MWR instrument is currently being developed to address these science questions for the Juno mission. As part of a deep space mission aboard a solar-powered spacecraft, MWR is designed to be compact, lightweight, and low power. The entire MWR instrument consists of six individual radiometer channels with approximately 4% bandwidth at 0.6, 1.25,2.6,5.2, 10,22 GHz operating in direct detection mode. Each radiometer channel has up to 80 dB of gain with a noise figure of several dB. The highest frequency channel uses a corrugated feedhorn and waveguide transmission lines, whereas all other channels use highly phase stable coaxial cables and either patch array or waveguide slot array antennas. Slot waveguide array antennas were chosen for the low loss at the next three highest frequencies and patch array antennas were implemented due to the mass constraint at the two lowest frequencies. The six radiometer channels receive their voltage supplies and control lines from an electronics unit that also provides the instrument communication interface to the Juno spacecraft. For calibration purposes each receiver has integrated noise diodes, a Dicke switch, and temperature sensors near each component that contributes to the noise figure. In addition, multiple sensors will be placed along the RF transmission lines and the antennas in order to measure temperature gradients. All antennas and RF transmission lines must withstand low temperatures and the harsh radiation environment surrounding Jupiter; the receivers and control electronics are protected by a radiation-shielding enclosure on the Juno spacecraft that also provides for a benign and stable operating temperature environment. This paper will focus on the concept of the MWR instrument and will present results of one breadboard receiver channel.

receiver↗
NASA NTRS2018

Implications of Whole-Disc DSCOVR EPIC Spectral Observations for Estimating Earth's Spectral Reflectivity Based on Low-Earth-Orbiting and Geostationary Observations

Earth's reflectivity is among the key parameters of climate research. National Aeronautics and Space Administration (NASA)'s Earth Polychromatic Imaging Camera (EPIC) onboard National Oceanic and Atmospheric Administration (NOAA)'s Deep Space Climate Observatory (DSCOVR) spacecraft provides spectral reflectance of the entire sunlit Earth in the near backscattering direction every 65 to 110 min. Unlike EPIC, sensors onboard the Earth Orbiting Satellites (EOS) sample reflectance over swaths at a specific local solar time (LST) or over a fixed area. Such intrinsic sampling limits result in an apparent Earth's reflectivity. We generated spectral reflectance over sampling areas using EPIC data. The difference between the EPIC and EOS estimates is an uncertainty in Earth's reflectivity. We developed an Earth Reflector Type Index (ERTI) to discriminate between major Earth atmosphere components: clouds, cloud-free ocean, bare and vegetated land. Temporal variations in Earth's reflectivity are mostly determined by clouds. The sampling area of EOS sensors may not be sufficient to represent cloud variability, resulting in biased estimates. Taking EPIC reflectivity as a reference, low-earth-orbiting-measurements at the sensor-specific LST tend to overestimate EPIC values by 0.8% to 8%. Biases in geostationary orbiting approximations due to a limited sampling area are between −0.7% and 12%. Analyses of ERTI-based Earth component reflectivity indicate that the disagreement between EPIC and EOS estimates depends on the sampling area, observation time and vary between −10% and 23%.

DSCOVR↗
NASA NTRS2023

Increasing Diversity on Spacecraft Mission Teams Reduces Risk

Many of the NASA spacecraft sent into Earth orbit and throughout the Solar System are competitively selected and led by principal investigators (PIs). These senior scientists are responsible for directing research activities, managing the use of funds, and reporting to the funding agency. Additionally, the European Space Agency (ESA) has competitively selected missions that are led by ESA-nominated project scientists and include instruments that are overseen by PIs. Usually, the PIs and project scientists form a science working team that provides scientific leadership of the mission. These NASA and ESA PI roles require a deep understanding of the mission’s scientific goals and engineering design aspects. And, importantly, these roles also require skills in team management.

Kathleen E. Mandt↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-ofthe-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Hesham ElAbd↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-of-the-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Deep-learning↗