Gotowa bibliografia na temat „MsMut1”

Mycobacterium smegmatis MutT1 (MsMutT1) is a sanitation enzyme made up of an N-terminal Nudix hydrolase domain and a C-terminal domain resembling a histidine phosphatase. It has been established that the action of MutT1 on 8-oxo-dGTP, 8-oxo-GTP and diadenosine polyphosphates is modulated by intermolecular interactions. In order to further explore this and to elucidate the structural basis of its differential action on 8-oxo-NTPs and unsubstituted NTPs, the crystal structures of complexes of MsMutT1 with 8-oxo-dGTP, GMPPNP and GMPPCP have been determined. Replacement soaking was used in order to ensure that the complexes were isomorphous to one another. Analysis of the structural data led to the elucidation of a relationship between the arrangements of molecules observed in the crystals, molecular plasticity and the action of the enzyme on nucleotides. The dominant mode of arrangement involving a head-to-tail sequence predominantly leads to the generation of NDPs. The other mode of packing arrangement appears to preferentially generate NMPs. This work also provides interesting insights into the dependence of enzyme action on the conformation of the ligand. The possibility of modulating the enzyme action through differences in intermolecular interactions and ligand conformations makes MsMutT1 a versatile enzyme.

With the rapid development of machine learning technology, how to use machine learning technology to empower the manufacturing industry has become a research hotspot. In order to solve the problem of product quality classification in a small sample data and imbalanced data environment, this paper proposes a data generation model called MSMOTE-GAN, which is based on Mahalanobis Synthetic Minority Oversampling Technology (MSMOTE) and Generative Adversarial Network (GAN). Among them, MSMOTE is proposed to solve the problem of the sample biased to the majority class expanded by methods such as GAN in a sample imbalanced environment. Based on the traditional SMOTE method, the sample distance measurement method is modified from Euclidean distance to Mahalanobis distance, taking into account the correlation between attributes and the influence of dimensions on the sample distance. In the data generation model, MSMOTE is used to balance the positive and negative samples in the data. GAN generates fake data with the same distribution as the original data based on a balanced data set and expands the sample size to solve the problems of overfitting and insufficient model expression ability that occur when the sample size is too small. The quality classification framework of water heater liner based on the data generation model and Random Forest is constructed, and the process of the quality classification of water heater liner under the environment of small sample data and imbalanced data is fully described. This paper compares the MSMOTE-GAN model, Bootstrap, and tableGAN on the water heater liner production line data set and the public data set. The experimental result shows that the expanded data set of the MSMOTE-GAN model can effectively improve the performance of the classification model.

Possible maintainer lines were selected from CC XXXII and crossed onto cytoplasmically male-sterile plants. Complete male sterility was maintained in both the F₁ and BC₁ generations of 46.4% of the lines. Four cultivars with maintainer genotypes that were in both normal and msm1 cytoplasm were intercrossed using the male-sterile forms as the female parents. All F₁'s were completely male-sterile. Restoration of male fertility by 22 lines selected from CC XXXII was shown in each case to be due to a single dominant gene. In some lines, restoration was influenced by environment and genetic background. Partial restoration was observed in cultivars in the World Collection and lines selected from CC XXXII. Partial restoration appeared to be due to several genes that were subject to environmental influence. Accumulation of some of these genes increased the amount of restoration. There was no evidence that cytoplasmic factors were passed through the pollen. Twenty-two F₁ hybrids were produced by crossing restorer lines onto male-sterile msm1 lines. The 22 hybrids, their 44 parental restorer and maintainer lines and six check cultivars were grown in a four-replication yield trial. Total yield, 1000-seed weight and hectaliter weight were measured for each plot. All the F₁ hybrids outyielded their midparent values and 17 of the hybrids outyielded their high parents. Half of the F₁'s outyielded the high check cultivar, which yielded about 9,130 kg/ha. Twenty-one F₁'s had greater 1000-seed weights than their midparent values while only 11 F₁'s had greater 1000-seed weights than their high parents. The high check cultivar had the greatest 1000-seed weight, 49.0 gm. The hybrids with the greatest 1000-seed weights were not the hybrids with the greatest yields. Eighteen of the F₁'s had greater hectaliter weights than the midparent values, but only seven had greater hectaliter weights than their high parents. The high check cultivar had the greatest hectaliter weight, 75 kg. The hybrids with the greatest hectaliter weights were not the highest yielding hybrids.

Maintenance of the genomic integrity of the cell is crucial for the survival and successful propagation of an organism. However, this integrity is under continuous threat from DNA-damaging agents. In addition, errors in replication and transcription, resulting in the incorporation of inappropriate bases and hence mutations, disrupt the genomic integrity. Therefore, cells have developed a number of DNA-repair and error avoidance mechanisms to maintain the genomic integrity. The genome of pathogenic mycobacteria, including Mycobacterium tuberculosis, is more prone to damage, as they are constantly exposed to DNA-damaging agents produced by the host macrophage which it inhabits. In addition, mycobacterial genomes, being rich in G+C content, are more susceptible to cytosine deamination and guanine oxidation. Therefore, various DNA-repair and error avoidance mechanisms for maintaining the genomic integrity are extremely important for mycobacteria. This laboratory has been actively contributing towards the national and international structural biology efforts on mycobacterial proteins involved in maintenance of the genomic integrity. The author’s contribution to this effort has been concerned with the base excision repair enzyme, uracil-DNA N-glycosylase (Ung), and nucleotide pool sanitization enzyme, MutT1 from mycobacteria. A brief overview of the available literature on the structural and biochemical studies of these two proteins is provided in the introductory chapter. Uracil in DNA can occur either due to deamination of cytosine within DNA or due to misincorporation of dUTP during replication. Uracil-DNA glycosylase is one of the important enzymes involved in the base excision repair (BER) pathway, which removes uracil form both single- and double-stranded DNA and hence avoids consequent mutations. 8-oxo-dGTP and 8-oxo-GTP are formed in the nucleotide pool as a consequence of oxidation of guanosine nucleotides. MutT proteins, sanitize the nucleotide pool and hence avoid the error due to incorporation of 8-oxo-dGTP and 8-oxo-GTP, during replication and transcription, respectively. Structural studies on uracil-DNA glycosylases from various sources including mycobacteria have been extensive, while those on MutT proteins have been less extensive. Uracil-DNA glycosylase is the first enzyme of the base excision repair pathway that removes uracil from both single- and double-stranded DNA by cleaving the bond between uracil and deoxyribose. UNG/Ung proteins are inhibited by a well known proteinaceous inhibitor, uracil-DNA glycosylase inhibitor (Ugi), which is encoded by the phages PBS-I and PBS-II as part of their defense mechanism against host Ung. Ugi has been well characterized biochemically and structurally and it has been extensively used in structural studies of UNG/Ung as it mimics the DNA bound to the enzyme. Apart from Ugi, UNG/Ungs are also inhibited to various extents by uracil, one of the products of the enzymatic reaction, and some of its analogs and derivatives. Although inhibition by these small molecules has been well studied biochemically, the modes of their binding and interactions with the enzyme have not been extensively explored. Structures of free UNG/Ung from many sources and their complexes with Ugi have been reported. The structure of the complex with oligonucleotides, however, is known only for the human, E. coli and Herpes simplex virus 1 (HSV1) enzymes. Of these, the oligonucleotide bound to HSV1 does not contain uracil. The structures of complexes of UNG/Ung with uracil, uracil analogs and a few of its derivatives have also been reported. Earlier analyses of the relevant structures revealed concerted conformational changes in the enzyme, leading to closing of the active-site cleft consequent to the binding of DNA containing uracil. Subsequently, it was demonstrated that Ung is a two-domain enzyme and that the domains close in on the bound DNA containing uracil. The mere presence of free uracil in the active site of the enzyme does not lead to the closure of the active site. The native structure of M. tuberculosis Ung (MtUng) with a citrate molecule bound in the active site and the structure of its complex with Ugi have previously been reported from this laboratory. The near atomic resolution structures of and thermodynamic data on complexes of MtUng with uracil and its derivatives and new crystal forms of the free enzyme, are reported here. A detailed structural examination of these high-resolution structures and the thermodynamic data have, among other things, led to striking insights into conformational selection on DNA binding and the modes of MtUng-ligand interactions. MutT proeins, which belong to Nudix hydrolase superfamily, are known to sanitize the nucleotide pool and hence avoid the error due to incorporation of non-canonical nucleotides, specifically 8-oxo-dGTP and 8-oxo-GTP, during replication and transcription, respectively. An antimutator role of MutT proteins have been established from a series of studies involving those from human, E. coli, M. tuberculosis and others sources. As indicated in the introductory chapter, mycobacterial MutT1 and MutT2 have been shown to have an 8-oxo-guanosine triphosphatase activity. Interestingly, unlike E. coli MutT (EcMutT) and human MutT homologue 1 (HsMTH1), M. tuberculosis MutT1 (MtMutT1) has been shown to hydrolyse 8-oxo-GTP and 8-oxo-dGTP to corresponding nucleoside diphosphates but not to nucleoside monophosphates at normal substrate and enzyme concentrations. To understand the basis of this novel and unusual activity and substrate specificity, it was desirable to carry out structural studies on MtMutT1. However, on account of problems with the expression of MtMutT1, Mycobacterium smegmatis MutT1 (MsMutT1) was chosen for structural studies. Preliminary studies on MsMutT1 suggested the presence of an N-terminal Nudix hydrolase domain (MsMutT1-NTD) corresponding to the single-domain EcMutT and a C-terminal histidine phosphatase domain (MsMutT1-CTD) in the protein. The presence of a phosphatase domain in MsMutT1 in addition to a Nudix hydrolase domain of the type that constitutes EcMutT and HsMTH1, was intriguing. Subsequently, detailed crystallographic studies of MsMutT1 and its complexes, along with complementary biochemical studies were carried out. The protein appears to be an enzyme in which the binding sites are formed primarily by intermolecular interactions of the type that bring the Nudix domain of one molecule and the phosphatase domain of a neighbouring molecule into close proximity. The enzyme is capable of hydrolysing 8-oxo-GTP and 8-oxo-dGTP into the corresponding nucleoside diphosphates and monophosphates, simultaneously, sequentially or both. A detailed examination of the crystal structures leads to a proposal as to how this is achieved. Diadenosine polyphosphates (ApnA, n=2–6), particularly Ap4A, are involved in several important physiological processes. The substantial sequence identity of the Nudix hydrolase domain (domain 1) of MsMutT1 with a known Ap4A hydrolase suggested that MsMutT1 could also hydrolyse diadenosine polyphosphates. Biochemical experiments yielded results in conformity with this suggestion, with Ap4A as the best among the substrates. ATP is a product in all experiments; small amounts of ADP were also observed in the experiments involving Ap4A and Ap6A. Hydrolysis was inhibited by fluoride ions in all cases. The mechanism of action and its inhibition in relation to ApnA were explored through the X-ray analysis of the crystals of the MsMutT1 complexes with Ap5A, Ap4A, ATP and ATP.MgF3. The aggregation pattern of molecules in these crystals is similar to that found in a majority of MsMutT1-NTP crystals. Substrate molecules occupy the same site in two of them. ATP occupies this site as well as another site at an intermolecular interface in the third crystal. The protein-ligand interactions observed in these crystal structures lead to an explanation of the molecular mechanism of hydrolysis of ApnA by MsMutT1. The crystals of the ATP.MgF3 complex exhibit a new packing arrangement. The structure of the complex provides an explanation for the fluoride inhibition of the activity of the enzyme. It would thus appear that MutT1 has a major role involving the hydrolysis of diadenosine polyphosphates, which could be elucidated at the molecular level.

碩士
國立交通大學
電控工程研究所
102
This thesis investigates the fault-tolerant output tracking issues for a class of nonlinear uncertain system using a combined scheme. The combined scheme consists of integral sliding mode control (ISMC) technique and control allocation (CA) idea. It is shown that the tracking performance suffering allowable actuator faults can be totally predicted from a preselected dynamics which can be selected according to the system requirements and provides a design degree of freedom. With this scheme, the control structure under different allowable faulty situations remains the same whenever the information of fault is available, so that the complexity of the fault-tolerant controller design is greatly reduced. The analytic results are then applied to the direct yaw moment control of a four-wheel driven electric vehicle. It is shown that the presented fault-tolerant scheme not only can achieve the yaw rate tracking task even when the allowable in-wheel actuators break down, but the tracking performance of allowable faulty situations is also close to that of the preselected error dynamic. Simulation results clearly demonstrate the benefits of the proposed fault-tolerant scheme.

The ability to express and purify soluble protein in significant amounts is a prerequisite for the structure analysis of biological macromolecules by spectroscopic techniques. Nuclear magnetic resonance (NMR) and electron paramagnetic resonance (EPR) spectroscopies are powerful biophysical techniques which are widely used in structural biology. This thesis focuses on the use of different fusion constructs and tagging strategies to produce samples for subsequent NMR and EPR measurements. Following the general introduction of Chapter 1, Chapter 2 explores N-terminal fusions based on the nucleotide sequence of the T7 gene 10 which translates into the hexapeptide MASMTG. A systematic comparison of the expression levels with and without MASMTG tag was conducted for five different proteins (E. coli aspartate/glutamate-binding protein (GBP), green fluorescent protein (GFP), MutT Homolog 1 (MTH1), dengue virus type 2 NS2B-NS3 protease (DENp) and Methanosarcina barkeri pyrrolysyl-tRNA synthetase (nbCRS)) in both a cell-free protein synthesis setup and in vivo in E. coli. The expression yields of DENp, GFP and nbCRS were greatly enhanced by the MASMTG tag, barely changed for GBP and decreased for MTH1. This result shows that the N-terminal fusion with a tag from a protein known to express in very high yields can indeed enhance the expression yields for some proteins even if they are already codonoptimized for expression in E. coli. Chapter 3 describes the development of an efficient and inexpensive strategy for site-specific paramagnetic tagging of oligonucleotides, which allowed measurements of pseudocontact shifts (PCS) in the DNA using lanthanide ion tags. The strategy relies on commercially available oligonucleotides synthesized with a phosphorothioate group. HPLC conditions were developed to separate the two phosphorothioate diastereomers and their configurations determined by an enzymatic assay with snake venom phosphodiesterase. The new lanthanide-binding tag C10 was attached by alkylating the phosphorothioate group. PCS measurements were carried out following hybridization with the complementary DNA strand to form DNA duplexes. Although the PCSs were relatively small, they confirmed the site-specific attachment of the tag. Larger PCSs were observed for the SP than the RP diastereomer and good correlations were observed between backcalculated and experimental PCSs, in particular for the SP-phosphorothioate oligonucleotide, indicating that this tagging approach delivers reliable long-range structural information. Chapter 4 describes the preparation of a homeodomain-DNA complex with three different types of spin labels for double electron–electron resonance (DEER) measurements. A Gd3+ tag was introduced into the homeodomain by copper-catalyzed click reaction with a genetically encoded unnatural amino acid (p-azidophenylalanine) and an EDTA-Mn2+ tag was introduced by reaction with Cys39. With a nitroxide tag attached to a phosphorothioate group in the DNA, DEER measurements determined the distances between the three labels in the triple-tagged homeodomain-DNA complex. The experimentally determined Mn2+−nitroxide and Gd3+−Mn2+ distance distributions agreed well with the distances predicted from the NMR structure of the complex, whereas the calculated Gd3+−nitroxide distance was ∼0.5 nm longer than the experimental one. This study demonstrated the potential of three different spin labels to obtain three independent distance restraints in a single sample. Chapter 5 describes experiments for the site-specific incorporation of the unnatural amino acids Boc-lysine and TMS-lysine into proteins using the Methanosarcina mazei pyrrolysine-tRNA synthetase (PylRS) mutants Y384F/Y306A and Y384F/Y306G/I405R in E. coli BL21 (DE3) and E. coli B95-DA. While Boc-lysine could readily be incorporated, the experiments to incorporate TMS-lysine were unsuccessful. Chapter 6 describes strategies explored to produce uniformly 15N-labelled MARCKS peptide.In vivo expression of this peptide had proven notoriously difficult but was successfully achieved by fusion with a trigger-factor–ubiquitin (TF-Ub) construct, which can be cleaved with a ubiquitinase to release the free peptide. 15N-HSQC spectra were recorded of the peptide in complex with the N60D mutant of calmodulin (CaM) loaded with calcium, and chemical shift changes and paramagnetic relaxation enhancements (PRE) detected in the presence of paramagnetic lanthanide ions confirmed specific binding to CaM and interactions with the N-terminal domain of CaM. This establishes the basis for future structural analysis of the binding mode of the MARCKS peptide to CaM. An intein strategy was unsuccessful for the expression of the MARCKS peptide, but the system successfully produced tag-free PylRS and allowed its purification in soluble form. The purified PylRS was inactive in the in-house cell-free protein synthesis (CFPS) system, indicating that the well-known problems with the activity of this enzyme are not associated with the presence of commonly used purification tags.

The paper analyses the problem of person re-identification accuracy in distributed video surveillance systems by using various datasets for training convolutional neural networks (CNN). After analysis we constructed large joint image dataset of people consisting of CUHK02, CUHK03, Market-1501, DukeMTMC-ReID, MSMT17 and our collected PolReID. PolReID includes 52035 images for 657 people Experimental results on assessing re-identification accuracy based on the main metrics Rank and mAP are presented. The research was carried out for the most widely used CNNs in re-identification, such as ResNet-50, DenseNet121 and PCB. We show that the constructed large dataset allowed us to improve Rank1, mAP for all test sets.

In higher education institutions (HEI), the ability to predict student grades as an early warning system is one of the important areas that gained attention to improve educational outcomes. Over the years, machine learning techniques have facilitated and successfully addressed student grade prediction for identifying the potentially weak students in a particular course. However, dealing with an imbalanced multiclass classification dataset is challenging due to biased results towards predicting the minority class. Therefore, this chapter proposes a method that can increase the classification performance by using a modified synthetic minority oversampling technique and feature selection (MSMOTE-FS). The experiments tested the proposed method's effectiveness by utilizing four oversampling techniques and six standard classification algorithms. This finding indicated that the proposed method gives promising results to improve the accuracy in multiclass classification of student grade prediction.

Model fine-tuning is a widely used transfer learning approach in person Re-identification (ReID) applications, which fine-tuning a pre-trained feature extraction model into the target scenario instead of training a model from scratch. It is challenging due to the significant variations inside the target scenario, e.g., different camera viewpoint, illumination changes, and occlusion. These variations result in a gap between the distribution of each mini-batch and the distribution of the whole dataset when using mini-batch training. In this paper, we study model fine-tuning from the perspective of the aggregation and utilization of the global information of the dataset when using mini-batch training. Specifically, we introduce a novel network structure called Batch-related Convolutional Cell (BConv-Cell), which progressively collects the global information of the dataset into a latent state and uses this latent state to rectify the extracted feature. Based on BConv-Cells, we further proposed the Progressive Transfer Learning (PTL) method to facilitate the model fine-tuning process by joint training the BConv-Cells and the pre-trained ReID model. Empirical experiments show that our proposal can improve the performance of the ReID model greatly on MSMT17, Market-1501, CUHK03 and DukeMTMC-reID datasets. The code will be released later on at \url{https://github.com/ZJULearning/PTL}

Recent works show that mean-teaching is an effective framework for unsupervised domain adaptive person re-identification. However, existing methods perform contrastive learning on selected samples between teacher and student networks, which is sensitive to noises in pseudo labels and neglects the relationship among most samples. Moreover, these methods are not effective in cooperation of different teacher networks. To handle these issues, this paper proposes a Graph Consistency based Mean-Teaching (GCMT) method with constructing the Graph Consistency Constraint (GCC) between teacher and student networks. Specifically, given unlabeled training images, we apply teacher networks to extract corresponding features and further construct a teacher graph for each teacher network to describe the similarity relationships among training images. To boost the representation learning, different teacher graphs are fused to provide the supervise signal for optimizing student networks. GCMT fuses similarity relationships predicted by different teacher networks as supervision and effectively optimizes student networks with more sample relationships involved. Experiments on three datasets, i.e., Market-1501, DukeMTMCreID, and MSMT17, show that proposed GCMT outperforms state-of-the-art methods by clear margin. Specially, GCMT even outperforms the previous method that uses a deeper backbone. Experimental results also show that GCMT can effectively boost the performance with multiple teacher and student networks. Our code is available at https://github.com/liu-xb/GCMT .