Shigeto Seno

Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer

Importance Application of deep learning algorithms to whole-slide pathology images can potentially improve diagnostic accuracy and efficiency. Objective Assess the performance of automated deep learning algorithms at detecting metastases in hematoxylin and eosin–stained tissue sections of lymph nodes of women with breast cancer and compare it with pathologists’ diagnoses in a diagnostic setting. Design, Setting, and Participants Researcher challenge competition (CAMELYON16) to develop automated solutions for detecting lymph node metastases (November 2015-November 2016). A training data set of whole-slide images from 2 centers in the Netherlands with (n = 110) and without (n = 160) nodal metastases verified by immunohistochemical staining were provided to challenge participants to build algorithms. Algorithm performance was evaluated in an independent test set of 129 whole-slide images (49 with and 80 without metastases). The same test set of corresponding glass slides was also evaluated by a panel of 11 pathologists with time constraint (WTC) from the Netherlands to ascertain likelihood of nodal metastases for each slide in a flexible 2-hour session, simulating routine pathology workflow, and by 1 pathologist without time constraint (WOTC). Exposures Deep learning algorithms submitted as part of a challenge competition or pathologist interpretation. Main Outcomes and Measures The presence of specific metastatic foci and the absence vs presence of lymph node metastasis in a slide or image using receiver operating characteristic curve analysis. The 11 pathologists participating in the simulation exercise rated their diagnostic confidence as definitely normal, probably normal, equivocal, probably tumor, or definitely tumor. Results The area under the receiver operating characteristic curve (AUC) for the algorithms ranged from 0.556 to 0.994. The top-performing algorithm achieved a lesion-level, true-positive fraction comparable with that of the pathologist WOTC (72.4% [95% CI, 64.3%-80.4%]) at a mean of 0.0125 false-positives per normal whole-slide image. For the whole-slide image classification task, the best algorithm (AUC, 0.994 [95% CI, 0.983-0.999]) performed significantly better than the pathologists WTC in a diagnostic simulation (mean AUC, 0.810 [range, 0.738-0.884]; P < .001). The top 5 algorithms had a mean AUC that was comparable with the pathologist interpreting the slides in the absence of time constraints (mean AUC, 0.960 [range, 0.923-0.994] for the top 5 algorithms vs 0.966 [95% CI, 0.927-0.998] for the pathologist WOTC). Conclusions and Relevance In the setting of a challenge competition, some deep learning algorithms achieved better diagnostic performance than a panel of 11 pathologists participating in a simulation exercise designed to mimic routine pathology workflow; algorithm performance was comparable with an expert pathologist interpreting whole-slide images without time constraints. Whether this approach has clinical utility will require evaluation in a clinical setting.

Proinflammatory cytokine interleukin-1β suppresses cold-induced thermogenesis in adipocytes

In this study, we investigated the effects of interleukin-1β (IL-1β), a typical proinflammatory cytokine on the β-adrenoreceptor-stimulated induction of uncoupling protein 1 (UCP1) expression in adipocytes. IL-1β mRNA expression levels were upregulated in white adipose tissues of obese mice and in RAW264.7 macrophages under conditions designed to mimic obese adipose tissue. Isoproterenol-stimulated induction of UCP1 mRNA expression was significantly inhibited in C3H10T1/2 adipocytes by conditioned medium from lipopolysaccharide (LPS)-stimulated RAW264.7 macrophages in comparison with control conditioned medium. This inhibition was significantly attenuated in the presence of recombinant IL-1 receptor antagonist and IL-1β antibody, suggesting that activated macrophage-derived IL-1β is an important cytokine for inhibition of β-adrenoreceptor-stimulated UCP1 induction in adipocytes. IL-1β suppressed isoproterenol-induced UCP1 mRNA expression in C3H10T1/2 adipocytes, and this effect was partially but significantly abrogated by inhibition of extracellular signal-regulated kinase (ERK). IL-1β also suppressed the isoproterenol-induced activation of the UCP1 promoter and transcription factors binding to the cAMP response element. Moreover, intraperitoneal administration of IL-1β suppressed cold-induced UCP1 expression in adipose tissues. These findings suggest that IL-1β upregulated in obese adipose tissues suppresses β-adrenoreceptor-stimulated induction of UCP1 expression through ERK activation in adipocytes.

Multilevel comparative analysis of the contributions of genome reduction and heat shock to the Escherichia coli transcriptome

BackgroundBoth large deletions in genome and heat shock stress would lead to alterations in the gene expression profile; however, whether there is any potential linkage between these disturbances to the transcriptome have not been discovered. Here, the relationship between the genomic and environmental contributions to the transcriptome was analyzed by comparing the transcriptomes of the bacterium Escherichia coli (strain MG1655 and its extensive genomic deletion derivative, MDS42) grown in regular and transient heat shock conditions.ResultsThe transcriptome analysis showed the following: (i) there was a reorganization of the transcriptome in accordance with preferred chromosomal periodicity upon genomic or heat shock perturbation; (ii) there was a considerable overlap between the perturbed regulatory networks and the categories enriched for differentially expressed genes (DEGs) following genome reduction and heat shock; (iii) the genes sensitive to genome reduction tended to be located close to genomic scars, and some were also highly responsive to heat shock; and (iv) the genomic and environmental contributions to the transcriptome displayed not only a positive correlation but also a negatively compensated relationship (i.e., antagonistic epistasis).ConclusionThe contributions of genome reduction and heat shock to the Escherichia coli transcriptome were evaluated at multiple levels. The observations of overlapping perturbed networks, directional similarity in transcriptional changes, positive correlation and epistatic nature linked the two contributions and suggest somehow a crosstalk guiding transcriptional reorganization in response to both genetic and environmental disturbances in bacterium E. coli.

INFERENCE OF S-SYSTEM MODELS OF GENE REGULATORY NETWORKS USING IMMUNE ALGORITHM

The S-system model is one of the nonlinear differential equation models of gene regulatory networks, and it can describe various dynamics of the relationships among genes. If we successfully infer rigorous S-system model parameters that describe a target gene regulatory network, we can simulate gene expressions mathematically. However, the problem of finding an optimal S-system model parameter is too complex to be solved analytically. Thus, some heuristic search methods that offer approximate solutions are needed for reducing the computational time. In previous studies, several heuristic search methods such as Genetic Algorithms (GAs) have been applied to the parameter search of the S-system model. However, they have not achieved enough estimation accuracy. One of the conceivable reasons is that the mechanisms to escape local optima. We applied an Immune Algorithm (IA) to search for the S-system parameters. IA is also a heuristic search method, which is inspired by the biological mechanism of acquired immunity. Compared to GA, IA is able to search large solution space, thereby avoiding local optima, and have multiple candidates of the solutions. These features work well for searching the S-system model. Actually, our algorithm showed higher performance than GA for both simulation and real data analyses.

Correlation between genome reduction and bacterial growth

Genome reduction by removing dispensable genomic sequences in bacteria is commonly used in both fundamental and applied studies to determine the minimal genetic requirements for a living system or to develop highly efficient bioreactors. Nevertheless, whether and how the accumulative loss of dispensable genomic sequences disturbs bacterial growth remains unclear. To investigate the relationship between genome reduction and growth, a series of Escherichia coli strains carrying genomes reduced in a stepwise manner were used. Intensive growth analyses revealed that the accumulation of multiple genomic deletions caused decreases in the exponential growth rate and the saturated cell density in a deletion-length-dependent manner as well as gradual changes in the patterns of growth dynamics, regardless of the growth media. Accordingly, a perspective growth model linking genome evolution to genome engineering was proposed. This study provides the first demonstration of a quantitative connection between genomic sequence and bacterial growth, indicating that growth rate is potentially associated with dispensable genomic sequences.

The hepatokine FGF21 is crucial for peroxisome proliferator-activated receptor-α agonist-induced amelioration of metabolic disorders in obese mice

Obesity causes excess fat accumulation in white adipose tissues (WAT) and also in other insulin-responsive organs such as the skeletal muscle, increasing the risk for insulin resistance, which can lead to obesity-related metabolic disorders. Peroxisome proliferator-activated receptor-α (PPARα) is a master regulator of fatty acid oxidation whose activator is known to improve hyperlipidemia. However, the molecular mechanisms underlying PPARα activator-mediated reduction in adiposity and improvement of metabolic disorders are largely unknown. In this study we investigated the effects of PPARα agonist (fenofibrate) on glucose metabolism dysfunction in obese mice. Fenofibrate treatment reduced adiposity and attenuated obesity-induced dysfunctions of glucose metabolism in obese mice fed a high-fat diet. However, fenofibrate treatment did not improve glucose metabolism in lipodystrophic A-Zip/F1 mice, suggesting that adipose tissue is important for the fenofibrate-mediated amelioration of glucose metabolism, although skeletal muscle actions could not be completely excluded. Moreover, we investigated the role of the hepatokine fibroblast growth factor 21 (FGF21), which regulates energy metabolism in adipose tissue. In WAT of WT mice, but not of FGF21-deficient mice, fenofibrate enhanced the expression of genes related to brown adipocyte functions, such as Ucp1, Pgc1a, and Cpt1b. Fenofibrate increased energy expenditure and attenuated obesity, whole body insulin resistance, and adipocyte dysfunctions in WAT in high-fat-diet-fed WT mice but not in FGF21-deficient mice. These findings indicate that FGF21 is crucial for the fenofibrate-mediated improvement of whole body glucose metabolism in obese mice via the amelioration of WAT dysfunctions.

Directed evolution of cell size in Escherichia coli.

BackgroundIn bacteria, cell size affects chromosome replication, the assembly of division machinery, cell wall synthesis, membrane synthesis and ultimately growth rate. In addition, cell size can also be a target for Darwinian evolution for protection from predators. This strong coupling of cell size and growth, however, could lead to the introduction of growth defects after size evolution. An important question remains: can bacterial cell size change and/or evolve without imposing a growth burden?ResultsThe directed evolution of particular cell sizes, without a growth burden, was tested with a laboratory Escherichia coli strain. Cells of defined size ranges were collected by a cell sorter and were subsequently cultured. This selection-propagation cycle was repeated, and significant changes in cell size were detected within 400 generations. In addition, the width of the size distribution was altered. The changes in cell size were unaccompanied by a growth burden. Whole genome sequencing revealed that only a few mutations in genes related to membrane synthesis conferred the size evolution.ConclusionsIn conclusion, bacterial cell size could evolve, through a few mutations, without growth reduction. The size evolution without growth reduction suggests a rapid evolutionary change to diverse cell sizes in bacterial survival strategies.

Perfect Hamming code with a hash table for faster genome mapping

BackgroundWith the advent of next-generation sequencers, the growing demands to map short DNA sequences to a genome have promoted the development of fast algorithms and tools. The tools commonly used today are based on either a hash table or the suffix array/Burrow–Wheeler transform. These algorithms are the best suited to finding the genome position of exactly matching short reads. However, they have limited capacity to handle the mismatches. To find n-mismatches, they requires O(2n) times the computation time of exact matches. Therefore, acceleration techniques are required.ResultsWe propose a hash-based method for genome mapping that reduces the number of hash references for finding mismatches without increasing the size of the hash table. The method regards DNA subsequences as words on Galois extension field GF(22) and each word is encoded to a code word of a perfect Hamming code. The perfect Hamming code defines equivalence classes of DNA subsequences. Each equivalence class includes subsequence whose corresponding words on GF(22) are encoded to a corresponding code word. The code word is used as a hash key to store these subsequences in a hash table. Specifically, it reduces by about 70% the number of hash keys necessary for searching the genome positions of all 2-mismatches of 21-base-long DNA subsequence.ConclusionsThe paper shows perfect hamming code can reduce the number of hash references for hash-based genome mapping. As the computation time to calculate code words is far shorter than a hash reference, our method is effective to reduce the computation time to map short DNA sequences to genome. The amount of data that DNA sequencers generate continues to increase and more accurate genome mappings are required. Thus our method will be a key technology to develop faster genome mapping software.

Evolutionary Consequence of a Trade-Off between Growth and Maintenance along with Ribosomal Damages.

Microorganisms in nature are constantly subjected to a limited availability of resources and experience repeated starvation and nutrition. Therefore, microbial life may evolve for both growth fitness and sustainability. By contrast, experimental evolution, as a powerful approach to investigate microbial evolutionary strategies, often targets the increased growth fitness in controlled, steady-state conditions. Here, we address evolutionary changes balanced between growth and maintenance while taking nutritional fluctuations into account. We performed a 290-day-long evolution experiment with a histidine-requiring Escherichia coli strain that encountered repeated histidine-rich and histidine-starved conditions. The cells that experienced seven rounds of starvation and re-feed grew more sustainably under prolonged starvation but dramatically lost growth fitness under rich conditions. The improved sustainability arose from the evolved capability to use a trace amount of histidine for cell propagation. The reduced growth rate was attributed to mutations genetically disturbing the translation machinery, that is, the ribosome, ultimately slowing protein translation. This study provides the experimental demonstration of slow growth accompanied by an enhanced affinity to resources as an evolutionary adaptation to oscillated environments and verifies that it is possible to evolve for reduced growth fitness. Growth economics favored for population increase under extreme resource limitations is most likely a common survival strategy adopted by natural microbes.

An estimation method for a cellular-state-specific gene regulatory network along tree-structured gene expression profiles

BACKGROUND Identifying the differences between gene regulatory networks under varying biological conditions or external stimuli is an important challenge in systems biology. Several methods have been developed to reverse-engineer a cellular system, called a gene regulatory network, from gene expression profiles in order to understand transcriptomic behavior under various conditions of interest. Conventional methods infer the gene regulatory network independently from each of the multiple gene expression profiles under varying conditions to find the important regulatory relations for understanding cellular behavior. However, the inferred networks with conventional methods include a large number of misleading relations, and the accuracy of the inference is low. This is because conventional methods do not consider other related conditions, and the results of conventional methods include considerable noise due to the limited number of observation points in each expression profile of interest. RESULTS We propose a more accurate method for estimating key gene regulatory networks for understanding cellular behavior under various conditions. Our method utilizes multiple gene expression profiles that compose a tree structure under varying conditions. The root represents the original cellular state, and the leaves represent the changed cellular states under various conditions. By using this tree-structured gene expression profiles, our method more powerfully estimates the networks that are key to understanding the cellular behavior of interest under varying conditions. CONCLUSION We confirmed that the proposed method in cell differentiation was more rigorous than the conventional method. The results show that our assumptions as to which relations are unimportant for understanding the differences of cellular states in cell differentiation are appropriate, and that our method can infer more accurately the core networks of the cell types.