Toxicogenomics : an approach to identify the mechanistic insights of toxicity including gene based biomarkers with the exposure of volatile organic compounds(VOCs) to human leukemia cells

Sarma, Sailendra Nath University of Science and Technology 2011 국내박사

Computational Design of Peptides as Detectors, Drugs and Biomaterials

Sarma, Sudeep ProQuest Dissertations & Theses North Carolina Sta 2023 해외박사(DDOD)

Computational design of peptides that can (a) recognize/bind to specific protein interfaces and (b) self-assemble into nano-scale architectures such as β-sheet-based fibrils (amyloid) have potential applications in healthcare and advanced biomaterial fabrication. This dissertation focusses on novel computational protocols and their applications to discover peptides that bind to protein targets and form amyloid structures. The key computational steps involved in these protocols are Monte-Carlo based peptide design algorithms (a peptide binding design algorithm, PepBD and a peptideassembly design algorithm , PepAD) to identify potential protein-binding peptides and self-assembling peptides, respectively. Additionally, atomistic molecular dynamics (MD) simulations are used to assess the stability of the designed peptide fibrils and the binding free energy of the peptide-protein complexes. Discontinuous molecular dynamics (DMD) simulation in conjunction with a coarse-grained model, PRIME20, is employed specifically for designing self-assembling peptides to reveal kinetic pathways for fibril formation.Our first study uses our PepBD algorithm and atomistic MD simulations to design peptides that can bind to the SARS-CoV-2 virus. Peptides capable of binding to the Spike-RBD of the Wuhan-Hu-1 strain were identified. In experimental evaluation, three of the five peptides that were synthesized bound to the Spike-RBD of the Wuhan-Hu-1 strain with dissociation constants in the micromolar range, but none of the peptides could outcompete the ACE2:RBD interactions. One peptide, P4, also bound to the SARS-CoV-2 RBD of Kappa-B.1.617.1 and Delta-B.1.617.2 with micromolar affinity. These results demonstrate our ability to design peptides that can recognize the broad spectrum of SARS-CoV-2 RBD variants.Next, we utilized our computational protocol to design 10-mer peptide inhibitors that block Clostridioides difficile toxin A in intestinal cells. We identified peptides that bind to the catalytic site of the Toxin A glucosyltransferase domain (GTD). Two of our in-silicodesigned peptides (NPA and NPB) exhibited lower binding free energies when bound to the TcdA GTD than a reference peptide, RP. In vitro experiments on human jejunum cells confirmed the toxinneutralizing properties of RP and NPA, but the efficacy of NPB was relatively low.Subsequently, we utilized our computational protocol to design 8-mer peptide inhibitors that block C. diff. toxin A in colon cells. (The 10-mer peptides that we designed previously for C. diff. TcdA neutralized C. diff. TcdA toxicity in small intestinal cells but showed no effect in the colon cells). Here, we developed 8-mer peptide inhibitors that block toxin A in both small intestinal cells and colon epithelial cells. Importantly, the designed peptide, SA1, demonstrated neutralization properties against toxin A toxicity in both the small intestine (SI) and colon; it bound toxin A with affinity binding constant KD= 56 ± 29.8 nM.Next, we developed a peptide assembly design (PepAD) algorithm to design peptides that form amyloid-like structures. DMD simulations were employed to reveal the kinetic pathway of fibril formation taken by the in-silicodiscovered peptides. We focused our efforts on designing peptides that form antiparallel amyloid-like structures, specifically the Class 8 cross-β spine described by Sawaya et al. Twelve 7-mer peptides capable of self-assembling into the desired structure were identified. DMD simulations revealed that eight of these peptides spontaneously form amyloid fibrils. Experimental tests confirmed the formation of antiparallel β-sheets at concentrations between 0.2 mM and 10 mM at room temperature in water.

Efficient closed-loop optimal control of petroleum reservoirs under uncertainty

Sarma, Pallav Stanford University 2006 해외박사(DDOD)

This work discusses a closed-loop approach for efficient realtime production optimization of petroleum reservoirs that consists of three key elements---adjoint models for efficient parameter and control gradient calculation, polynomial chaos expansions for efficient uncertainty propagation, and Karhunen-Loeve (K-L) expansions and Bayesian inversion theory for efficient realtime model updating (history matching). The control gradients provided by the adjoint solution are used by a gradient-based optimization algorithm to determine optimal control settings, while the parameter gradients are used for model updating. We also investigate an adjoint construction procedure that makes it relatively easy to create the adjoint and is applicable to any level of implicitness of the forward model. Polynomial chaos expansions provide optimal encapsulation of information contained in the input random fields and output random variables. This approach allows the forward model to be used as a black box but is much faster than standard Monte Carlo techniques. The K-L representation of input random fields allows for the direct application of adjoint techniques for history matching and uncertainty propagation algorithms while assuring that the two-point geostatistics of the reservoir description are maintained. We further extend the basic closed-loop algorithms discussed above to address two important issues. The first concerns handling non-linear path inequality constraints during optimization, which are quite difficult to maintain with existing optimal control algorithms. We propose an approximate feasible direction algorithm combined with a feasible line-search to satisfy such constraints efficiently. The second issue concerns the Karhunen-Loeve expansion. It is computationally very expensive and impractical for large-scale simulation models, and since it only preserves two-point statistics of the input random field, it may not always be suitable for arbitrary non-Gaussian random fields. We use Kernel Principal Component Analysis (PCA) to address these issues efficiently. This approach is much more efficient, preserves high-order statistics of the random field, and is differentiable, meaning that gradient-based methods (and adjoints) can still be utilized with this representation. The benefits and efficiency of the overall closed-loop approach are demonstrated through realtime optimizations of net present value (NPV) for synthetic and real reservoirs under waterflood subject to production constraints and uncertain reservoir description.

Developmental dynamics of piriform cortex

Sarma, Amy Akella Yale University 2011 해외박사(DDOD)

The piriform cortex (PCX) is a trilaminar paleocortex that is of interest for its role in odor coding and as a model for studying general principles of cortical sensory processing. While the structure of the mature PCX has been well characterized, its development is poorly understood. Notably, the kinetics as well as the cellular and morphological basis of the postnatal events that shape the PCX remain unknown. We followed the cellular fates of early- versus late-born cells in layer II of the anterior PCX, with a focus on the molecular maturation of pyramidal cells and the kinetics of their differentiation. We showed that: (1) early-born pyramidal cells differentiate more rapidly than late-born cells; and (2) the position of pyramidal cells within the thickness of layer II determines the kinetics of their molecular maturation. We then examined the postnatal development of cortical lamination and showed that the establishment of inhibitory networks in the PCX proceeds through an increase in the density of inhibitory synapses despite a decrease in the number of interneurons. Together, our results provide a more comprehensive view of the postnatal development of the anterior PCX and reveal both similarities and differences in the development of this paleocortex versus the neocortex.

Fuzzy discrete multicriteria cost optimization of steel structures using genetic algorithm

Sarma, Kamal Chandra The Ohio State University 2001 해외박사(DDOD)

A great majority of optimization works on steel structures deal with minimum weight design. But in reality minimum weight design is not necessarily a minimum cost design. The price of commercially available rolled steel shapes in the market does not necessarily depend only on its weight but also on some other factors including demands and grades. In cost optimization, however, additional difficulties are encountered. They include the definition of cost function and uncertainties and fuzziness involved in determining the cost parameters. As a result only a small fraction of the structural optimization papers published deal with the minimization of the cost. In this work a fuzzy discrete multicriteria initial cost optimization model has been developed by considering three criteria: (1) minimum cost, (2) minimum weight, and (3) minimum number of section types. This discrete cost optimization procedure is preceded by a fuzzy genetic continuous variable minimum weight design as a preliminary design. In this design the uncertainty or fuzziness of the AISC code based design constraints are considered. Furthermore, the fuzzy multicriteria initial cost optimization model is extended to perform a life cycle cost optimization of steel structures. For optimization of large steel structures more than 99% of computation time is spent on the minimum weight design using the fuzzy Genetic Algorithm. Sequential processing of this optimization work in a single processor is very inefficient resulting in huge numbers of page faults and cache misses even in a supercomputer like Origin 2000. In this work two bi-level parallel processing algorithms are developed by using (1) data parallel procedure using OpenMP API at inner level and (2) message passing distributed parallel processing procedure by using function calls from MPI message passing library at outer level. Significant performance enhancement has been obtained in comparison to the traditional data parallel or distributed message passing parallel processing.

Parental Pressure for Academic Success in India

Sarma, Arti Arizona State University 2014 해외박사(DDOD)

Academic achievement among Asians has been widely recognized in the literature, but the costs of this success may be tied to significant mental health consequences. Three samples of undergraduate students in India were recruited from cities such as Chennai, Kerala, and Delhi totaling 608 (303 male, 301 females). Both online and in class recruitment occurred. There were three main purposes of this study: 1) to construct a quantitative measure of parental pressure, 2) to evaluate whether self-esteem was a potential buffer of the negative impacts of parental pressure and academic stress, and 3) to understand better the factors impacting suicidality among adolescents in India by testing a path model of possible predictors suggested by the literature. Prevalence data of suicidal ideation and attempt history were also collected. Reporting on their experience over the past six months, 14.5% (n = 82) of the participants endorsed suicidal ideation and 12.3% (n = 69) of the participants admitted to having deliberately attempted to hurt or kill themselves. Five constructs were explored in this study: parental pressure, academic stress, depression, suicidality, and self-esteem. The Parental Pressure for Success Scale, designed for this study, was used to measure parental pressure. The Educational Stress Scale-Adolescents was used to measure academic stress. The Center for Epidemiological Studies-Depression scale was used to measure depressive symptomology. Two items from the Youth Self-Report Checklist were used as a measure of suicidality in the past six months. The Rosenberg Self-esteem Scale was used to measure global self-esteem. Preliminary support for the reliability and validity of the Parental Pressure for Success Scale was found. While self-esteem was not a significant moderator in this study, it was a predictor of both stress and depression. Results of the path analysis indicated that parental pressure predicted academic stress, stress predicted depression, and depression predicted suicidality. Parental pressure indirectly predicted suicidality through academic stress and depression. Results were discussed in the context of cultural influences on study findings such as the central role of parents in the family unit, the impact of cultural valuing of education, collectivistic society, and the Hindu concept of dharma, or duty.

Adaption of the Minnesota Multiphasic Personality Inventory-2 to Latvia

Sarma, Zinta Mara University of Minnesota 2005 해외박사(DDOD)

The purpose of the present investigation was to adapt the MMPI-2 for use in Latvia. The project encompassed two phases. Phase I consisted of a multi-step process of translating the MMPI-2 to Latvian. Phase II consisted of three separate investigations that examined evidence of reliability and validity of the Latvian MMPI-2. In the first study, 26 Latvian-English bilingual individuals completed both the Latvian and English version of the MMPI-2 in a test-retest format; resultant correlation data suggested that the Latvian MMPI-2 was equivalent to the U.S. version. A second study compared the MMPI-2 data from 181 female and 76 male Latvian undergraduate students with data previously collected from a sample of U.S. university students. The Latvian MMPI-2 scales demonstrated adequate internal consistency, and scale intercorrelations were similar to those observed in the U.S. Scale level analyses revealed many similarities in the basic validity, clinical and content scale score configurations. Female Latvian student scale scores were slightly higher on the D and Hy clinical scales when compared to their U.S. counterparts, and both female and male Latvian students' results revealed slight elevations above the U.S. student group data on several of the content scales. The F infrequency scale was also elevated for both Latvian females and males. None of the scales were elevated above the clinically significant T-score of 65. Principal components analysis revealed a four-component structure that paralleled the structure typically found in U.S. populations. In Study 3, MMPI-2 profiles of a group of 56 Latvian psychiatric inpatients were compared to those of the Latvian female and male students. Multiple scales were elevated well above the Latvian student data and were clinically significant. The mean MMPI-2 scale score configuration of 26 Latvian inpatients with a diagnosis of schizophrenia was highly similar to that of a U.S. group with the same diagnosis. Implications of the results were reviewed; the minor differences observed in the Latvian student profiles were discussed in terms of possible cultural differences between Latvia and the U.S.

Algorithms for large graphs

Das Sarma, Atish Georgia Institute of Technology 2010 해외박사(DDOD)

This thesis explores algorithms for massive graphs. We explore efficient graph algorithms under standard models of processing such large data sets. The main graph problem considered in this thesis is performing random walks efficiently. Random walks are fundamental across all of computer science ranging from theory, mathematics, distributed computing and web algorithms. Several applications include search, data mining, ranking (such as PageRank), measuring similarity between web pages, maintaining connectivity in P2P networks etc. The work in this thesis has developed the fastest algorithms for performing random walks in the streaming model, and in distributed networks, breaking past the linear-time barrier presented by the sequential nature of random walks. This work improves upon techniques that have been used for decades in both theory as well as practice. In follow up work on the streaming model presented in this thesis, we consider the problem of graph partitioning. Several offline and few streaming/online algorithms have been suggested for partitioning an input graph into two parts, to approximate the conductance of the graph. While many of these techniques are impractical at the scale of the massive web graphs, some may be implementable. However, when dealing with several hundred million nodes and tens of billions of edges, visualizing one global cut becomes very difficult. What can one say from a partition that separates the graph into two humongous sets of nodes? In this work, we consider the problem of finding what we call cut projections. Given a (possibly small) subset of nodes from the graph, the objective is to partition the set of nodes such that this projects onto an induced cut of small conductance on the entire graph. The hope is that the bounds now strongly depend on the size of the specified set of nodes and only weakly on the size of the entire graph. We show how such cut projections can be obtained on regular graphs when the subset of nodes is chosen at random. While our theorems do not hold when these restrictions are omitted, the technique itself is likely to work well in practice. In the distributed computing model, we again focus on the problem of performing random walks efficiently. Our results on this model work only on undirected networks (which is usually the case, since the communication paths are bidirectional). In follow up work under the same congest model of distributed networks, we further improve the algorithm for performing several random walks. Further, we show a lower bound that suggests that a single random walk requires O( ℓ/logℓ ) rounds. Therefore, barring the diameter term, our approach for performing a single random walk is near-optimal. Our algorithm also introduces several new techniques and random walk properties that may be of independent interest. Further, we show how our algorithms for performing random walks efficiently can be used as subroutines for efficient distributed algorithms for two key properties: random spanning trees, and estimating mixing times. We show the fastest known distributed algorithms for both sampling a random spanning tree (which is useful in applications such as routing and generating sparsifiers), and estimating mixing times (which is crucial to understanding the connectivity of the network, a primary concern in peer to peer systems). Finally, in the online framework of sketch based algorithms, we study the problem of estimating graph distances efficiently. Our objective is to be able to answer distance queries between a pair of nodes in real time. Since the standard shortest path algorithms are expensive, our approach moves the time-consuming shortest-path computation offline, and at query time only looks up precomputed values and performs simple and fast computations on these precomputed values. More specifically, during the offline phase we compute and store a small sketch for each node in the graph, and at query-time we look up the sketches of the source and destination nodes and perform a simple computation using these two sketches to estimate the distance. (Abstract shortened by UMI.).

Numerical simulations of two-way coupling effects in a particle-laden turbulent pipe flow, and, Evaluation of the equilibrium Eulerian approach for the evolution of particle concentration in isotropic turbulence

Rani, Sarma Laxminarasimha University of Illinois at Urbana-Champaign 2002 해외박사(DDOD)

Turbulence modulation in a fully developed downward pipe flow by dense particles at a Reynolds number of 360 based on friction velocity and pipe diameter is studied using direct numerical simulations. Particles are smaller than the Kolmogorov scale of turbulence and are treated using the Lagrangian approach. The effects of varying particle response time, volume fraction and settling velocity on fluid turbulence are investigated. In the absence of gravity, variation of either the particle volume fraction or the response time has negligible effects on the fluid streamwise mean velocity profile. However, an increase in either the volume fraction or the response time augments the streamwise RMS velocities and attenuates the radial and azimuthal RMS velocities. Particles with positive settling velocity lead to a marginal increase in the mean streamwise fluid velocities and a substantial increase in the streamwise RMS velocities. The fluid radial and azimuthal RMS velocities, in the presence of gravity, are higher than those for the no-gravity case. Turbulence augmentation at the smaller dissipative scales can be seen in the energy spectra at certain radial locations. The longitudinal energy spectra show greatest turbulence augmentation close to the pipe center and the pipe wall. In case of azimuthal energy spectra, augmentation at medium and high wavenumbers is seen close to the pipe center. Particles with non-zero settling velocity provide some augmentation in the azimuthal energy spectra near the wall. In the presence of gravity, due to the reverse cascade, turbulence augmentation at low wavenumbers is observed. Particle collisions reduce the degree of particle migration to the wall. Their effects on fluid turbulence are also presented. The second part of this thesis concerns the application of the equilibrium Eulerian approach to study particle preferential concentration and settling velocity in isotropic turbulence. The equilibrium Eulerian approach is extended to evolve the particle concentration for varying particle response time and still-fluid settling velocity. Over the entire range of particle parameters considered, there is good agreement between the Eulerian and the Lagrangian statistics. The equilibrium Eulerian approach tends to overpredict preferential concentration, compared to the Lagrangian particles, at higher response times.

Managing uncertain data

Das Sarma, Anish Stanford University 2010 해외박사(DDOD)

The ubiquity of uncertain data in modern-day applications (such as information extraction, data integration, sensor and RFID networks, and scientific experiments) has resulted in a growing need for techniques to deal with such data. This thesis addresses challenges in managing uncertain data in a principled, usable, and scalable fashion. We identify and explore a fundamental tension between usability and expressiveness in models for representing uncertain data. We propose a space of models for representing uncertain data, place the models in an expressiveness hierarchy, and study how the models relate to each other in terms of closure properties. We also address important problems of uniqueness testing, equivalence checking, minimization, and approximation in our space of models. For a representative model in our space (called URM), we study database design theory: We provide a sound and complete axiomatization of functional dependencies (FDs) for URM data, describe lossless decompositions, and give algorithms and complexity results for testing, finding, and inferring FDs. To address the usability-expressiveness tradeoff, we show that by adding lineage (provenance) to the URM model, we obtain a complete (intuitively, a fully expressive) data model, which we call the Uncertainty-Lineage Database (ULDB) model. We study properties of ULDBs including membership, extraction, and minimization. We develop techniques for query processing over ULDBs and show that lineage can be exploited for efficient confidence computation in ULDBs. Then, we present an extension to ULDBs that allows a seamless incorporation of data modifications and a lightweight versioning capability. Finally, we look at uncertain data management in the context of data integration. Data integration systems offer a uniform interface to a set of data sources. Despite recent progress, setting up and maintaining a data integration application still requires significant up-front effort. We present a completely self-configuring data integration system based on a probabilistic framework. The system produces high-quality query answers with no human intervention.