Treasures @ UT Dallas
Not a member yet
7697 research outputs found
Sort by
Development of a High-throughput Screening Method to Identify Chloride-sensitive Variants of the Fluorescent Protein mPapaya
No full textEngineering genetically encoded chloride-sensitive fluorescent biosensors to understand the
behavior of chloride in biological systems is a promising area of research because these biosensors
have the ability to target specific locations and the ability to express within specific cells in
transgenic organisms while selectively enabling them to determine the behavior of chloride within
a cell, both qualitatively and quantitatively. The GFP-derived fluorescent proteins are used in this
engineering process as their unique structure enables them to act as anion recognition domain,
mainly through the hydrophobic interactions and the formation of hydrogen bonds while the
autocatalytic chromophore contributes to generating an optical output as a transducer. Despite the
unique properties, these fluorescent proteins exist with certain limitations and therefore need to be
modified through mutagenesis to get the desirable features to enable them to function as
biosensors. To accomplish this process, detailed structural analysis of the protein structure
including chloride binding coordination sphere and chromophore environment is required as the
structural analysis will help to identify the sites for introducing mutations. After identifying the
potential sites which would foster the desirable features, site-saturation mutagenesis will be used
to generate the library. As the mutagenesis library contains a vast number of platforms of mutated
variants, including the variants in which our desired feature (chloride-sensitivity) is
enhanced/deprived or intact variants with no mutations, there should be a validated method for
library preparation, screening and then identifying the chloride-sensitive variants from the library
in high-throughput format.
This thesis will explain about the development of a workflow to screen chloride-responsive
variants in a high-throughput format while specifying every step, from selecting a vector,
designing primers, library preparation, screening, and sequencing, as all these steps have a major
impact on the protein engineering. We will use mPapaya, as the testing protein and describe a
detailed structural analysis to provide a broad picture on selecting positions for engineering the
chloride binding pocket in it. The process described in this thesis involves in the identification of
an appropriate mutation site through sequence alignment with avYFP-H148Q and the identified
site (H202/ p-position) is mutated using site-saturation mutagenesis where the chloride-sensitive
variants can be identified based on confidence interval graphs. Future considerations will be to use
this method to identify the mutated variants in other testing fluorescent proteins in our lab
Ultrasound Sensing of Lower-limb Skeletal Muscle and the Potential for Robotic Assistive Device Control
No full textAn estimated 11% of the world's population either faces difficulty with daily mobility or uses
assistive devices such as walkers, canes, crutches, and wheelchairs, reducing the quality of life and
impairing the independence of their living. Lower-limb robotic assistive devices such as prosthetic
limbs and exoskeletons hold a great promise to improve the quality of life of these individuals by
replicating most of the actions of healthy biological limbs. However, the clinical translation and
widespread adoption of these devices are still hindered by the lack of control intuitiveness and
continuous adaptation to user intentions. Ultrasound is a noninvasive sensing modality that can
access human neuromuscular information by measuring muscle activation and contraction. Hence,
the objective of the present dissertation was to investigate the feasibility of using the ultrasound
sensing modality as a wearable human-machine interface for lower-limb robotic assistive devices.
We hypothesize that human lower-limb muscle kinematics and kinetics, noninvasively tracked by
ultrasound sensing, can be used to estimate continuous lower-limb joint movements. We further
hypothesize that ultrasound sensing can be used as a sensing interface with a continuous volitional
control framework to integrate with lower-limb robotic assistive devices. In this dissertation, we:
1) develop a framework to extract muscle contraction features from ultrasound images demonstrate
their use for lower-limb joint motion estimation, 2) develop an ultrasound-based control
framework for continuous joint-level control of lower-limb assistive devices during various
steady-state and non-steady-state ambulation modes, and 3) improve the potential integration of
the ultrasound technology with existing lower-limb assistive devices by reducing the integrationlimiting parameters of the current ultrasound technology. The outcomes of this dissertation may
eventually enhance the clinical relevance of lower-limb robotic assistive devices and facilitate their
widespread adoption for the population with mobility problems
Common-mode Current and EMI Analysis for Wide Bandgap Device Based Power Converters
No full textWide bandgap devices such as Silicon Carbide (SiC) and Gallium Nitride (GaN) are facilitating
higher switching frequencies for power electronic converters. Their faster switching speed
combined with low on-state resistance helps in increasing the efficiency and power density of the
converters. However, high voltage and current slew rates (dv/dt and di/dt) associated with wide
bandgap devices also increase Electromagnetic Interference (EMI) which may cause
malfunctioning or failure of the converter itself or other sensitive circuits onboard. EMI issues are
further worsened when high switching speed and high frequency operation is used for high power
applications. Apart from hard switched wide bandgap power converters, dv/dt rates can also be
high in wide bandgap converters which offer Zero Voltage Switching (ZVS) resulting into high
Common-Mode (CM) currents through parasitic capacitances. High CM current leads to EMI.
Dual Active Bridge (DAB) converter is a DC-DC converter offering ZVS along with bi-directional
power transfer, isolation, and voltage matching. It finds applications in Electric Vehicles (EVs),
battery charging stations, Solid State Transformers (SSTs) etc. High dv/dt, high switching
frequency combined with multiple parasitic elements present in an isolated converter like DAB,
make EMI mitigation difficult.
This dissertation presents CM current and EMI analysis for single phase DAB converter.
Conventional and Neutral Point Clamp (NPC) based architectures are studied with regard to their
suitability to reduce CM current. Design and control variables in Active NPC (ANPC) - DAB
converter are used to reduce CM conducted as well as radiated EMI. While doing so, it is insured
that all possible modulation schemes used in DAB converters remain unaffected.
To further reduce CM current than what is possible by design and/or control, an Input Parallel
Output Parallel (IPOP) ANPC-DAB architecture is proposed for high power DAB converters. This
architecture uses power sharing and dv/dt cancellation to reduce CM current injection. Factors
causing nonideal dv/dt cancellations are discussed with their possible solutions.
Finally in a contrasting work, EMI is investigated as a noninvasive system monitoring tool. A low
cost radiated EMI sensing circuit is developed to detect dynamic current unbalancing in parallel
connected wide bandgap devices. Apart from its hardware validation for paralleled SiC MOSFETs,
possibilities to use it for EMI precompliance testing and online junction temperature monitoring
are also discussed
Stray Magnetic Flux Based Condition Monitoring Techniques for Permanent Magnet Synchronous Motors
No full textPermanent Magnet Synchronous Motors (PMSM) are widely used in various applications from
home appliances to electric vehicles due to its advantages like high efficiency, high power density
etc. The market share of this type of motor is increasing significantly in the recent decade. In order
to ensure their reliable operation in safety critical system, condition monitoring tools are essential.
The effective fault diagnosis and condition monitoring system reduces downtime and unplanned
maintenance which saves both time and money. Though several motor attributes like current,
vibration, temperature etc. can be used for diagnosis of health condition of an electric motor, stray
magnetic flux around the motor housing has rich information about the health status of PM motors
due to the use of high energy magnets. This dissertation presents condition monitoring techniques
which are developed to address three major issues in PMSM.
Stator fault is the most common electrical fault which begins as an inter-turn failure. It is essential
to identify this failure at its initial stage to prevent the catastrophic damage caused by stator fault.
It is shown that the fundamental component of stray magnetic flux may result in fault negative
error under certain operating conditions. To overcome this problem, third harmonic component is
analyzed comprehensively and proposed as a reliable fault metric throughout the entire range of
operation to detect and locate the fault.
Secondly, the common mechanical fault in electrical motors, bearing damage is studied. It is shown
that the magnetic field of PM motors exhibit asymmetry due to the manufacturing imperfections
of magnet. A stray magnetic flux-based bearing fault detection method is proposed by leveraging
this asymmetry. The proposed method is analyzed extensively in both simulations and
experiments. A comparative study between motor current based detection and the proposed
method is also performed.
In the last part of the research, a stray magnetic flux based the temperature estimation of permanent
magnet is proposed. The stray magnetic flux around the motor can be correlated to the magnet
temperature. However, it also is affected by the permeability variation of iron core and magnetic
field generated by the stator current. A compensation co-efficient is proposed to compensate the
mentioned effects and obtain the permanent magnet component of stray magnetic flux from the
measured data. Then the proposed scheme is adopted to estimate the temperature of magnets online
under dynamic operating conditions
Effects of Transcranial Magnetic Stimulation on Individual Functional Brain Networks
No full textTranscranial magnetic stimulation (TMS) has long been utilized as a tool to non-invasively study
the association between brain function and cognition and is increasingly being used as a
therapeutic intervention to modify brain function in an effort to treat disease. Towards these
goals, recent efforts have strived to target and modify large-scale functional brain networks. This
work is predicated on observations that the organization of the human functional brain network is
related to behavioral performance among individuals across the lifespan, and cognitive decline
that has been observed in health aging and in disease states. Large-scale brain networks consist
of nodes (brain areas) that vary in their functional characteristics. Nodes belong to distinct
subnetworks that represent functional brain systems related to cognition and sensory-motor
function. Functional brain network organization has been observed to differ in specific
topography and localization across individuals, which motivates efforts towards identifying
individual-specific stimulation targets to reduce group-level bias and inaccuracies. This
dissertation project utilized resting-state functional correlations (RSFC) to map the functional
brain networks of individuals and identify specific stimulation targets in a group of healthy
young adult participants (N = 17[9F], 18-30y). RSFC MRI was collected at baseline and 24
hours after completing a 5-day high-frequency rTMS protocol. For each individual participant,
two cortical stimulation targets with distinct functional and topological properties were
identified: the left angular gyrus (L. Ang.) and left middle frontal gyrus (L. MFG), with each
target serving as an active control condition for the other target. On-target stimulation of each
cortical target resulted in RSFC changes between the respective target node and its connections
in the network, whereas RSFC changes were less evident following off-target (control-site)
stimulation. RSFC changes were demonstrated to be related to the baseline RSFC strength and
Euclidean distance between each target and their respective network connections, and was also
related to on-target stimulation, demonstrating distal impacts of TMS that are mediated by both
functional and anatomical features of functionally connected brain regions. On-target TMS to the
L. Ang. decreased RSFC within the default-mode system (DMN) and on-target TMS to the L.
MFG decreased RSFC within the frontoparietal control system (FPN) as well as in the default-
mode system; off-target control stimulation had no impact on RSFC within the DMN. While
node-level and system-level correlations were modified as a result of TMS, system segregation,
an overall measure quantifying brain network organization was not impacted by TMS,
highlighting the potential resiliency of a segregated brain network in the face of TMS
perturbation. Notably, measures of episodic memory performance (but also fluid ability and
working memory) were not impacted by TMS and were not related to observed changes in
RSFC, despite previous reports that have provided evidence for these effects. This dissertation
provides evidence that individualized on-target TMS modifies RSFC in a target-specific manner
and is related to functional properties of the stimulated node within the brain network
Knowledge Extraction From Email Conversations and Its Application to Question Answering
No full textEmail communication is the exchange of messages between two or more people over the
internet using electronic devices. With billions of emails exchanged every day, extracting
knowledge from emails is beneficial to various email-based user applications. In any form
of communication, it is important to identify the participating users or entities and their
interactions throughout the conversation. Previous research on email processing has primarily focused on classification, searching, and intent detection. However, it has overlooked
studying the interaction between entities participating in an email conversation.
One of the tasks to capture the interaction is entity coreference resolution. End-to-end entity coreference resolution extracts entities and their references throughout the conversation.
Post extraction, it is important to use a knowledge representation format that preserves
and enriches the extracted knowledge. Knowledge graphs can assist in representing these
extracted intra-email interactions compactly. They can also enrich the knowledge by capturing inter-email interactions using a robust technique like matching entities across email
conversations. One of the main applications to use the extracted knowledge is question answering that focuses on the entities in a conversation. These tasks, when put together, paint
a simplistic yet holistic picture of a knowledge extraction pipeline for email conversations.
A deep joint learning framework is proposed for the novel task of entity coreference resolution
for email conversations. Two datasets were created during the framework’s development
process. These datasets were used to evaluate the task difficulty and identify the limitations
of the available solutions. The framework used the task of text classification for joint learning
to improve the scoring of text spans. This task was also used for incorporating singletons in
the result. The joint learning framework and singleton addition achieved an improvement of
4.87 and 5.26 F1 points on the two datasets, respectively.
A combination of automatic and manual methods was used to carry out relation extraction
parallel to entity coreference resolution. The extracted knowledge from the tasks was used to
create two knowledge graphs - one that contained the knowledge from the relation extraction
task and the other that contained knowledge extracted from both tasks. The knowledge
graph creation process used the NEPOMUK framework, which was created to simplify data
sharing across different user applications. Changes to the NEPOMUK framework have been
proposed for adding coreference knowledge to the graphs.
Lastly, previous work has investigated doing question answering using digital voice assistants.
However, this dissertation explores the novel task setting of doing question answering using
digital voice assistants for email conversations. The sub-task of entity resolution has been
identified as essential to the proposed formulation, and a dataset evaluating the same was
created using templates. A deep learning-based and two SPARQL template-based systems
were used for the evaluation process. Empirical results showed an increase of 3.91% in
the template-based system’s accuracy when coreference information was incorporated in
knowledge graphs. By laying a framework for the knowledge extraction pipeline, creating
open-source datasets and benchmarks for comparison, one can hope to advance research in
email processing
Exploiting Instance Similarity in Applications of Deep Learning in Bioinformatics
No full textPredicting the state or function of a biological organism from gross observation is a recurrent challenge in biology. As biological systems are too complex, identifying the rules and
principles from the observations is extremely hard. In recent years, machine learning has
been employed to deal with the complexity of biological data. Machine learning techniques
often rely on hand-crafted features. However, designing and leveraging such features are not
always obvious. Over the last decade, deep learning has emerged as a new area in machine
learning, revolutionized our understanding of biology. Unfortunately, the sparsity of training
data in some specific domains has restricted the use of deep learning. One way to overcome
this limitation is to utilize instance similarities. In many applications, however, incorporating
similarity information into deep learning models is not straightforward. Utilizing instance
similarity in biology has been the main motivation of this research.
In this dissertation, we have tackled exploiting the similarity of instances into deep learning
in two main applications: a) drug-target interaction prediction, and b) learning morphological similarity in histopathology images. In the first application, we introduce a deep
learning framework to predict drug-target interactions by learning the topological features
from a bipartite drug-target interaction graph. Furthermore, we exploit drug and protein
similarity information by extending our framework to learn from a semi-bipartite graph. We
show that our approach achieves state-of-the-art performance in predicting new drug-target
interactions. In the second application, we propose a novel deep metric learning methodology that learns morphological similarities in histopathology images. Thanks to a new
task and metric learning design, our approach performs without requiring any labeled data.
We demonstrate our approach learns more general purpose features than discriminative approaches, and therefore performs better in downstream tasks. We show that our framework
can be used as a backbone in a few tasks, i.e., image retrieval, identifying morphological and
biological correlation and transfer learning. Furthermore, We show our approach can also be
used to reduce the batch effect in histopathology domain. We also show that our approach’s
strength and unsupervised nature, make it a powerful tool for biological explorations and
discoveries
Cloud Detection and PM2.5 Estimation Using Machine Learning
No full textEarth observation (EO) is the gathering of information about the physical, chemical, and
biological systems of the planet via remote-sensing technologies, supplemented by Earthsurveying techniques, which encompasses the collection, analysis, and presentation of data.
Research on exploring effective methods for earth observation data analysis has increased
over the years because of the increasing amount of data generated by earth observation
systems, such as remote sensing imagery and weather radars. Researchers have therefore
taken an interest in machine learning, a technique that allows computer algorithms to learn
from samples. In general, the more comprehensive our training samples are, the better
the machine learning performance will be. This feature makes machine learning an ideal
approach for analyzing earth observation data. Particulate matter of fine size, such as
particulate matter 2.5 (PM2.5), poses a severe health risk to humans and is associated with
many different health problems. PM2.5 concentrations are influenced by factors such as
meteorological conditions, local population density, and the geographic context. As a result
of the large quantity of information provided by Earth observation, they become a valuable
tool for studying PM2.5. They are huge and come from different platforms, with different
spatial and temporal resolutions, and in different formats, which challenge the approaches
for PM2.5 studies.
This dissertation shows how machine learning methods can be used to address these challenges in three subtopics connected to modeling and estimation for PM2.5. Satellite-based
remote sensing products provide important variables that can be used to study regional and
global PM2.5, such as the Aerosol Optical Depth (AOD). Nevertheless, AOD products in
cloudy areas cannot be retrieved, and the quality of AOD data in nearby cloud areas cannot
be guaranteed.
Accordingly, the first study aims to detect cloud pixels based on remote sensing images. This
study investigates the cloud detection with a set of machine learning models on four subsets
of 88 Landsat8 images that have been carefully labelled by analysts. Four subsets of training
data are used to train 16 machine learning models with different input feature selections.
The performance of these models is then compared with that of the Fmask algorithm, which
is widely used for cloud detection. When testing on the 88 annotated images, the best
performance was observed with a model that incorporates unsupervised self-organizing map
(SOM) classification results among the input features. In comparison with Fmask4.0, the
model improves the correctness by 10.11% and reduces the cloud omission error by 6.39%.
Focusing on the other 8 independent validation images that were never sampled as part of
the model training, the model trained on the second largest training subset with additional
5 input features has the best overall performance. Compared with Fmask4.0, this model
improves the overall correctness by 3.26% and reduces the cloud omission error by 1.28%.
In the second study, high temporal resolution PM2.5 models are developed based on data from
weather radar systems and the meteorological data from the European Centre for MediumRange Weather Forecasts (ECMWF). A dataset covering the period from July 2019 to June
2021 was collected for model training, which included the Next Generation Weather Radar
(NEXRAD) retrieved from a repository on Amazon Web Services (AWS), meteorological
data from ECMWF, and the PM2.5 ground observations from 31 sensors deployed across
Dallas county, Collin county, and Tarrant county. The models are classified in groups to demonstrate the effectiveness of NEXRAD in high temporal PM2.5 modeling. The model
utilizing NEXRAD data achieves an 0.855 score of the correlation of determination (R2
),
while the model without NEXRAD has a 0.7 R2
for PM2.5.
The third study establishes a nationwide PM2.5 estimation model by using high temporal resolution AOD data from the GOES-16 geostationary satellite, meteorological variables from
ECMWF and a set of ancillary data from a variety of sources, which achieves 3.0µg/m3 and
5.8 µg/m3 as the value of mean absolute error (MAE) and root mean square error (RMSE).
The model performances are then further evaluated by time, elevation, soil order, population density, and lithology. The historical PM2.5 estimation surfaces are then reconstructed
and the PM2.5 surfaces during the period of California Santa Clara Unite (SCU) Lightning
Complex fires are demonstrated
Design and Validation of Switched Moving Boundary Modeling for Phase Change Thermal Energy Storage Systems
No full textThermal Energy Storage (TES) devices, which leverage the constant-temperature thermal
capacity of the latent heat of a Phase Change Material (PCM), provide benefits to a variety of thermal management systems by decoupling the absorption and rejection of thermal
energy. Control-oriented models are needed to predict the behavior of the TES to maximize
the capabilities and efficiency of the overall system and experimental validation is needed
to demonstration the validity of the simplifying assumptions used to produce these control-
oriented models. This thesis experimentally demonstrates the predictive capabilities of a
switched Moving Boundary (MB) model that captures the key dynamics of the TES with
significantly fewer states as compared to traditional approaches. A graph-based modeling
approach is used to model the heat flow through the TES and the moving boundary captures the time-varying liquid and solid regions of the TES. A Finite State Machine (FSM)
is used to switch between four different modes of operation based on the State-of-Charge
(SOC) of the TES. The switched MB approach is shown to have similar accuracy and lower
computational cost compared to traditional modeling approaches when predicting the SOC
of an experimental TES device
The Transition Metal Catalyzed Polymerization of Heteroallenes: Cholesteric Gels and Cyclopolymerizations
No full textFrom microscopic to macroscopic scale, the helix is a distinctive structure which can be found in
nature. Synthetically prepared helices impart a wide variety of applications such as chiral
separations, chiral sensing materials, optical materials, liquid crystalline materials, drug
application and asymmetric catalysis to mention but a few. Improved materials for these
applications by development of novel helical frameworks can unlock doors to a wide expansion in
the real world. Chapter 1 briefly discusses the synthetic helical polymers, more specifically
polycarbodiimides and polyisocyanates with their monomer synthesis, polymerization methods
developed over the years and their controlled chiral polymer backbone formations along with their
applications. Chapter 2 explains how the chirality in an achiral polymer can be induced using
copolymerization. In other words, how single-handed chiral segments of static helical polymers of
carbodiimides can be used to convert a racemic mixture of dynamic helical polymers of
polyisocyanates into adapting a chiral polymer backbone. Copolymers of isocyanates and
carbodiimides using CpTiCl2OCH2CF3 as the catalyst have been synthesized. Effective random
copolymerization can be observed in the resulting polymers which has improved the chirality and handedness of the otherwise achiral poly(hexylisocyanate). Chapter 3 focuses on inducing
chirality of the achiral racemic mixture of polyisocyanates using light cross-linking to lock in their
helical reversals along the polymer backbone. The liquid crystalline mesophases of these crosslinked polymers have been improved from nematic mesophases to cholesteric mesophases.
Chapter 4 discusses the synthesis and characterization of unique polymers of 1,2-dicarbodiimides
that possess a cyclic structure throughout the polymer backbone by tethering the two functional
groups close enough to promote their intramolecular reaction during the polymerization. New
cyclo-polycarbodiimide polymers have been synthesized and display significant differences when
compared to their acyclic counterparts