Yuki Saito

Bio: Yuki Saito is an academic researcher from University of Tokyo. The author has contributed to research in topics: Speech synthesis & Catalysis. The author has an hindex of 20, co-authored 102 publications receiving 1456 citations.

Extremotolerant tardigrade genome and improved radiotolerance of human cultured cells by tardigrade-unique protein

Abstract: Tardigrades, also known as water bears, are small aquatic animals. Some tardigrade species tolerate almost complete dehydration and exhibit extraordinary tolerance to various physical extremes in the dehydrated state. Here we determine a high-quality genome sequence of Ramazzottius varieornatus, one of the most stress-tolerant tardigrade species. Precise gene repertoire analyses reveal the presence of a small proportion (1.2% or less) of putative foreign genes, loss of gene pathways that promote stress damage, expansion of gene families related to ameliorating damage, and evolution and high expression of novel tardigrade-unique proteins. Minor changes in the gene expression profiles during dehydration and rehydration suggest constitutive expression of tolerance-related genes. Using human cultured cells, we demonstrate that a tardigrade-unique DNA-associating protein suppresses X-ray-induced DNA damage by ∼40% and improves radiotolerance. These findings indicate the relevance of tardigrade-unique proteins to tolerability and tardigrades could be a bountiful source of new protection genes and mechanisms.

Statistical Parametric Speech Synthesis Incorporating Generative Adversarial Networks

Abstract: A method for statistical parametric speech synthesis incorporating generative adversarial networks (GANs) is proposed. Although powerful deep neural networks techniques can be applied to artificially synthesize speech waveform, the synthetic speech quality is low compared with that of natural speech. One of the issues causing the quality degradation is an oversmoothing effect often observed in the generated speech parameters. A GAN introduced in this paper consists of two neural networks: a discriminator to distinguish natural and generated samples, and a generator to deceive the discriminator. In the proposed framework incorporating the GANs, the discriminator is trained to distinguish natural and generated speech parameters, while the acoustic models are trained to minimize the weighted sum of the conventional minimum generation loss and an adversarial loss for deceiving the discriminator. Since the objective of the GANs is to minimize the divergence (i.e., distribution difference) between the natural and generated speech parameters, the proposed method effectively alleviates the oversmoothing effect on the generated speech parameters. We evaluated the effectiveness for text-to-speech and voice conversion, and found that the proposed method can generate more natural spectral parameters and $F_0$ than conventional minimum generation error training algorithm regardless of its hyperparameter settings. Furthermore, we investigated the effect of the divergence of various GANs, and found that a Wasserstein GAN minimizing the Earth-Mover's distance works the best in terms of improving the synthetic speech quality.

Non-Parallel Voice Conversion Using Variational Autoencoders Conditioned by Phonetic Posteriorgrams and D-Vectors

Abstract: This paper proposes novel frameworks for non-parallel voice conversion (VC) using variational autoencoders (VAEs). Although conventional VAE-based VC models can be trained using non-parallel speech corpora with given speaker representations, phonetic contents of the converted speech tend to vanish because of an over-regularization issue often observed in latent variables of the VAEs. To overcome the issue, this paper proposes a VAE-based non-parallel VC conditioned by not only the speaker representations but also phonetic contents of speech represented as phonetic posteriorgrams (PPGs). Since the phonetic contents are given during the training, we can expect that the VC models effectively learn speaker-independent latent features of speech. Focusing on the point, this paper also extends the conventional VAE-based non-parallel VC to many-to-many VC that can convert arbitrary speakers' characteristics into another arbitrary speakers' ones. We investigate two methods to estimate speaker representations for speakers not included in speech corpora used for training VC models: 1) adapting conventional speaker codes, and 2) using d-vectors for the speaker representations. Experimental results demonstrate that 1) PPGs successfully improve both naturalness and speaker similarity of the converted speech, and 2) both speaker codes and d-vectors can be adopted to the VAE-based many-to-many non-parallel VC.

Transcription and Translation Products of the Cytolysin Gene psm-mec on the Mobile Genetic Element SCCmec Regulate Staphylococcus aureus Virulence

Abstract: The F region downstream of the mecI gene in the SCCmec element in hospital-associated methicillin-resistant Staphylococcus aureus (HA-MRSA) contains two bidirectionally overlapping open reading frames (ORFs), the fudoh ORF and the psm-mec ORF. The psm-mec ORF encodes a cytolysin, phenol-soluble modulin (PSM)-mec. Transformation of the F region into the Newman strain, which is a methicillin-sensitive S. aureus (MSSA) strain, or into the MW2 (USA400) and FRP3757 (USA300) strains, which are community-acquired MRSA (CA-MRSA) strains that lack the F region, attenuated their virulence in a mouse systemic infection model. Introducing the F region to these strains suppressed colony-spreading activity and PSMα production, and promoted biofilm formation. By producing mutations into the psm-mec ORF, we revealed that (i) both the transcription and translation products of the psm-mec ORF suppressed colony-spreading activity and promoted biofilm formation; and (ii) the transcription product of the psm-mec ORF, but not its translation product, decreased PSMα production. These findings suggest that both the psm-mec transcript, acting as a regulatory RNA, and the PSM-mec protein encoded by the gene on the mobile genetic element SCCmec regulate the virulence of Staphylococcus aureus.

Novel mechanisms of glucocorticoid resistance in childhood acute lymphoblastic leukemia.

Abstract: 2458 Despite dramatic improvements in treatment over the past 40 years, acute lymphoblastic leukemia (ALL) remains one of the most common causes of death from disease in childhood. Glucocorticoids are among the most effective agents used in the treatment of lymphoid malignancies, and patient response to treatment is an important determinant of long-term outcome in childhood ALL. In spite of its clinical significance, the molecular basis of glucocorticoid resistance is still poorly understood. The aim of this study was to develop an experimental model system to define clinically relevant mechanisms of glucocorticoid resistance in childhood ALL. An in vivo model of childhood ALL has been developed in our laboratory, using patient biopsies established as xenografts in immune-deficient nonobese diabetic severe-combined immunodeficient (NOD/SCID) mice. This model is highly representative of the human disease (Lock et al., Blood, 99: 4100-4108, 2002). The in vivo responses of these xenografts to the glucocorticoid dexamethasone (DEX) correlated significantly with patient outcome (p 1 μM) in xenografts from six patients, five of whom died of their disease. In contrast, four DEX-sensitive xenografts (IC50 values 2-fold in sensitive xenografts within 8 hours of treatment. In contrast, Bim induction was dramatically attenuated in DEX-resistant xenografts. These results have identified a clinically significant and novel mechanism of glucocorticoid resistance in childhood ALL, which occurs downstream of receptor-ligand interactions, but upstream of the signalling pathway resulting in Bim induction and apoptosis.

Human lymphoid and myeloid cell development in NOD/LtSz-scid IL2R gamma null mice engrafted with mobilized human hemopoietic stem cells.

Abstract: Ethical considerations constrain the in vivo study of human hemopoietic stem cells (HSC). To overcome this limitation, small animal models of human HSC engraftment have been used. We report the development and characterization of a new genetic stock of IL-2R common gamma-chain deficient NOD/LtSz-scid (NOD-scid IL2Rgamma(null)) mice and document their ability to support human mobilized blood HSC engraftment and multilineage differentiation. NOD-scid IL2Rgamma(null) mice are deficient in mature lymphocytes and NK cells, survive beyond 16 mo of age, and even after sublethal irradiation resist lymphoma development. Engraftment of NOD-scid IL2Rgamma(null) mice with human HSC generate 6-fold higher percentages of human CD45(+) cells in host bone marrow than with similarly treated NOD-scid mice. These human cells include B cells, NK cells, myeloid cells, plasmacytoid dendritic cells, and HSC. Spleens from engrafted NOD-scid IL2Rgamma(null) mice contain human Ig(+) B cells and lower numbers of human CD3(+) T cells. Coadministration of human Fc-IL7 fusion protein results in high percentages of human CD4(+)CD8(+) thymocytes as well human CD4(+)CD8(-) and CD4(-)CD8(+) peripheral blood and splenic T cells. De novo human T cell development in NOD-scid IL2Rgamma(null) mice was validated by 1) high levels of TCR excision circles, 2) complex TCRbeta repertoire diversity, and 3) proliferative responses to PHA and streptococcal superantigen, streptococcal pyrogenic exotoxin. Thus, NOD-scid IL2Rgamma(null) mice engrafted with human mobilized blood stem cells provide a new in vivo long-lived model of robust multilineage human HSC engraftment.

Humanized mice in translational biomedical research

Abstract: The culmination of decades of research on humanized mice is leading to advances in our understanding of human haematopoiesis, innate and adaptive immunity, autoimmunity, infectious diseases, cancer biology and regenerative medicine. In this Review, we discuss the development of these new generations of humanized mice, how they will facilitate translational research in several biomedical disciplines and approaches to overcome the remaining limitations of these models.

Multirate digital signal processing

Abstract: There has been a great deal of material in the area of discrete-time transforms that has been published in recent years. This book does an excellent job of presenting important aspects of such material in a clear manner. The book has 11 chapters and a very useful appendix. Seven of these chapters are essentially devoted to the Fourier series/transform, discrete Fourier transform, fast Fourier transform (FFT), and applications of the FFT in the area of spectral estimation. Chapters 8 through 10 deal with many other discrete-time transforms and algorithms to compute them. Of these transforms, the KarhunenLoeve, the discrete cosine, and the Walsh-Hadamard transform are perhaps the most well-known. A lucid discussion of number theoretic transforms i5 presented in Chapter 11. This reviewer feels that the authors have done a fine job of compiling the pertinent material and presenting it in a concise and clear manner. There are a number of problems at the end of each chapter, an appreciable number of which are challenging. The authors have included a comprehensive set of references at the end of the book. In brief, this book is a valuable contribution to the general areas of signal processing and communications. It can be used for a graduate level course in perhaps two ways. One would be to cover the first seven chapters in great detail. The other would be to cover the whole book by focussing on different topics in a selective manner. This book by Elliott and Rao is extremely useful to researchers/engineers who are working in the areas of signal processing and communications. It i s also an excellent reference book, and hence a valuable addition to one’s library