Search

Biological data may be separated into primary data, such as gene expression, and secondary data, such as pathways and protein-protein interactions. Methods using secondary data to enhance the analysis of primary data are promising, because secondary data have background information that is not included in primary data. In this study, we proposed an end-to-end framework to integrally handle secondary data to construct a classification model for primary data. We applied this framework to cancer prognosis prediction using gene expression data and a biological network. Cross-validation results indicated that our model achieved higher accuracy compared with a deep neural network model without background biological network information. Experiments conducted in patient groups by cancer type showed improvement in ROC-area under the curve for many groups. Visualizations of high accuracy cancer types identified contributing genes and pathways by enrichment analysis. Known biomarkers and novel biomarker candidates were identified through these experiments., 19 pages including supplementary materials

Drug repositioning holds great promise because it can reduce the time and cost of new drug development. While drug repositioning can omit various R&D processes, confirming pharmacological effects on biomolecules is essential for application to new diseases. Biomedical explainability in a drug repositioning model can support appropriate insights in subsequent in-depth studies. However, the validity of the XAI methodology is still under debate, and the effectiveness of XAI in drug repositioning prediction applications remains unclear. In this study, we propose GraphIX, an explainable drug repositioning framework using biological networks, and quantitatively evaluate its explainability. GraphIX first learns the network weights and node features using a graph neural network from known drug indication and knowledge graph that consists of three types of nodes (but not given node type information): disease, drug, and protein. Analysis of the post-learning features showed that node types that were not known to the model beforehand are distinguished through the learning process based on the graph structure. From the learned weights and features, GraphIX then predicts the disease-drug association and calculates the contribution values of the nodes located in the neighborhood of the predicted disease and drug. We hypothesized that the neighboring protein node to which the model gave a high contribution is important in understanding the actual pharmacological effects. Quantitative evaluation of the validity of protein nodes' contribution using a real-world database showed that the high contribution proteins shown by GraphIX are reasonable as a mechanism of drug action. GraphIX is a framework for evidence-based drug discovery that can present to users new disease-drug associations and identify the protein important for understanding its pharmacological effects from a large and complex knowledge base., add supplementary materials

Learning stable dynamics from observed time-series data is an essential problem in robotics, physical modeling, and systems biology. Many of these dynamics are represented as an inputs-output system to communicate with the external environment. In this study, we focus on input-output stable systems, exhibiting robustness against unexpected stimuli and noise. We propose a method to learn nonlinear systems guaranteeing the input-output stability. Our proposed method utilizes the differentiable projection onto the space satisfying the Hamilton-Jacobi inequality to realize the input-output stability. The problem of finding this projection can be formulated as a quadratic constraint quadratic programming problem, and we derive the particular solution analytically. Also, we apply our method to a toy bistable model and the task of training a benchmark generated from a glucose-insulin simulator. The results show that the nonlinear system with neural networks by our method achieves the input-output stability, unlike naive neural networks. Our code is available at https://github.com/clinfo/DeepIOStability., Comment: NeurIPS 2022

Early disease detection and prevention methods based on effective interventions are gaining attention. Machine learning technology has enabled precise disease prediction by capturing individual differences in multivariate data. Progress in precision medicine has revealed that substantial heterogeneity exists in health data at the individual level and that complex health factors are involved in the development of chronic diseases. However, it remains a challenge to identify individual physiological state changes in cross-disease onset processes because of the complex relationships among multiple biomarkers. Here, we present the health-disease phase diagram (HDPD), which represents a personal health state by visualizing the boundary values of multiple biomarkers that fluctuate early in the disease progression process. In HDPDs, future onset predictions are represented by perturbing multiple biomarker values while accounting for dependencies among variables. We constructed HDPDs for 11 non-communicable diseases (NCDs) from a longitudinal health checkup cohort of 3,238 individuals, comprising 3,215 measurement items and genetic data. Improvement of biomarker values to the non-onset region in HDPD significantly prevented future disease onset in 7 out of 11 NCDs. Our results demonstrate that HDPDs can represent individual physiological states in the onset process and be used as intervention goals for disease prevention.

We propose a new approach to SAT solving which solves SAT problems in vector spaces as a cost minimization problem of a non-negative differentiable cost function J^sat. In our approach, a solution, i.e., satisfying assignment, for a SAT problem in n variables is represented by a binary vector u in {0,1}^n that makes J^sat(u) zero. We search for such u in a vector space R^n by cost minimization, i.e., starting from an initial u_0 and minimizing J to zero while iteratively updating u by Newton's method. We implemented our approach as a matrix-based differential SAT solver MatSat. Although existing main-stream SAT solvers decide each bit of a solution assignment one by one, be they of conflict driven clause learning (CDCL) type or of stochastic local search (SLS) type, MatSat fundamentally differs from them in that it continuously approach a solution in a vector space. We conducted an experiment to measure the scalability of MatSat with random 3-SAT problems in which MatSat could find a solution up to n=10^5 variables. We also compared MatSat with four state-of-the-art SAT solvers including winners of SAT competition 2018 and SAT Race 2019 in terms of time for finding a solution, using a random benchmark set from SAT 2018 competition and an artificial random 3-SAT instance set. The result shows that MatSat comes in second in both test sets and outperforms all the CDCL type solvers.

[Purpose]The purpose of our study was to predict blood pressure variability from time-series data of blood pressure measured at home and data obtained through medical examination at a hospital. Previous studies have reported the blood pressure variability is a significant independent risk factor for cardiovascular disease. [Methods]We adopted standard deviation for a certain period and predicted variabilities and mean values of blood pressure for 4 weeks using multi-input multi-output deep neural networks. In designing the prediction model, we prepared a dataset from a clinical study. The dataset included past time-series data for blood pressure and medical examination data such as gender, age, and others. As evaluation metrics, we used the standard deviation ratio (SR) and the root-mean-square error (RMSE). Moreover, we used cross-validation as the evaluation method. [Results]The prediction performances of blood pressure variability and mean value after 1–4 weeks showed the SRs were “0.67” to “0.70”, the RMSEs were “5.04” to “6.65” mmHg, respectively. Additionally, our models were able to work for a participant with high variability in blood pressure values due to its multi-output nature. [Conclusion]The results of this study show that our models can predict blood pressure over 4 weeks. Our models work for an individual with high variability of blood pressure. Therefore, we consider that our prediction models are valuable for blood pressure management., Kojima, Ryosuke, Kario, Kazuomi, Okuno, Yasushi

金沢大学ナノマテリアル研究所 / 金沢大学理工研究域生命理工学系, As an alternative to petroleum-based compatibilizing agents, we developed lignin derivatives for compatibilizing agents of carbon fiber-reinforced plastics that have thermoplasticity. In this study, alkyl chains were introduced into alkali lignin at various ratios to optimize the compatibility of the lignin derivatives with both polypropylene and carbon fiber. The interfacial shear strength between the two materials was improved from 8.2 to 17.2 MPa by mixing with the optimized lignin derivative. The value is comparable to that achieved with a typical petroleum-based compatibilizing agent (18.3 MPa)., Embargo Period 6 months, This paper has supplementary information.

We introduce a new logic programming language T-PRISM based on tensor embeddings. Our embedding scheme is a modification of the distribution semantics in PRISM, one of the state-of-the-art probabilistic logic programming languages, by replacing distribution functions with multidimensional arrays, i.e., tensors. T-PRISM consists of two parts: logic programming part and numerical computation part. The former provides flexible and interpretable modeling at the level of first order logic, and the latter part provides scalable computation utilizing parallelization and hardware acceleration with GPUs. Combing these two parts provides a remarkably wide range of high-level declarative modeling from symbolic reasoning to deep learning. To embody this programming language, we also introduce a new semantics, termed tensorized semantics, which combines the traditional least model semantics in logic programming with the embeddings of tensors. In T-PRISM, we first derive a set of equations related to tensors from a given program using logical inference, i.e., Prolog execution in a symbolic space and then solve the derived equations in a continuous space by TensorFlow. Using our preliminary implementation of T-PRISM, we have successfully dealt with a wide range of modeling. We have succeeded in dealing with real large-scale data in the declarative modeling. This paper presents a DistMult model for knowledge graphs using the FB15k and WN18 datasets., Comment: In AAAI-19 Workshop on Network Interpretability for Deep Learning

Birds in the wild are difficult to localize, because their sizes tend to be small, they move swiftly, and they are often visually occluded. However, their location information is crucial for ethological studies on birds' behaviour. Recently, automating the process has been studied as a hot topic, where spatial sensors and sensor networks are commonly used. To avoid the visual occlusion problem, many studies focus on acoustic signal processing by applying microphone arrays and perform 1D azimuth localization through bird songs. In this study, we perform 2D sound source localization in the Cartesian coordinates using azimuths from multiple microphone arrays. To estimate the exact bird's location, we calculate the intersection points of these azimuth lines. Although this approach is simple and easy to be implemented, it has two main issues. One is that even small noise interference in azimuth values results in corrupting the localization data. This leads to a problem, where the intersection points between the azimuth lines do not intersect in one point for a single bird, but in several points. This proves difficulty in estimating the exact location of each bird. Especially in a far-field application, even small noise corruption leads to large localization errors. The other issue is that in the bird's natural habitat, elements such as leaves, grass and rivers are natural noise sources. It is difficult to extract the bird songs in such a noisy environment. We propose an algorithm involving statistic methods, sound feature analysis and machine learning. Based on this approach, a noise robust bird localization system has been established. We have performed numerous simulations to further understand the limitations of the system. Based on the results we have also derived the system's design guidelines, describing how the results change depending on the number of microphone arrays, signal-to-noise ratio, bird's distance from the devices, array's transfer function, type of the singing bird and specific parameter settings used in the algorithms. Such detailed guidelines support interested researchers in creating a similar system, which can contribute to ethological researches.

This thesis addresses scene analysis, which is essential for environmental monitoring and understanding. The main problems for scene analysis are twofold; extraction of attributes and reconstruction of scene structure from signal data obtained from sensors. To resolve them, we propose a scene analysis framework by combining unified event extraction and stochastic plan recognition. Unified event extraction considers the probabilistic relationship between attributes of a scene event to extract attributes more accurately than conventional methods dealing with attributes separately. Stochastic plan recognition structures the extracted events with back-ground knowledge, which yields plausible reconstruction of scene structure. This thesis also shows effectiveness of the proposed framework by introducing three applications: cooking recognition, bird song analysis, and web session log monitoring.

Exosomes are cell-derived nanovesicles (50–150 nm), which mediate intercellular communication, and are candidate therapeutic agents. However, inefficiency of exosomal message transfer, such as mRNA, and lack of methods to create designer exosomes have hampered their development into therapeutic interventions. Here, we report a set of EXOsomal transfer into cells (EXOtic) devices that enable efficient, customizable production of designer exosomes in engineered mammalian cells. These genetically encoded devices in exosome producer cells enhance exosome production, specific mRNA packaging, and delivery of the mRNA into the cytosol of target cells, enabling efficient cell-to-cell communication without the need to concentrate exosomes. Further, engineered producer cells implanted in living mice could consistently deliver cargo mRNA to the brain. Therapeutic catalase mRNA delivery by designer exosomes attenuated neurotoxicity and neuroinflammation in in vitro and in vivo models of Parkinson’s disease, indicating the potential usefulness of the EXOtic devices for RNA delivery-based therapeutic applications., Nature Communications, 9 (1), ISSN:2041-1723

Complementing enzymes in their native environment with either homogeneous or heterogeneous catalysts is challenging due to the sea of functionalities present within a cell. To supplement these efforts, artificial metalloenzymes are drawing attention as they combine attractive features of both homogeneous catalysts and enzymes. Herein we show that such hybrid catalysts consisting of a metal cofactor, a cell-penetrating module, and a protein scaffold are taken up into HEK-293T cells where they catalyze the uncaging of a hormone. This bioorthogonal reaction causes the upregulation of a gene circuit, which in turn leads to the expression of a nanoluc-luciferase. Relying on the biotin–streptavidin technology, variation of the biotinylated ruthenium complex: the biotinylated cell-penetrating poly(disulfide) ratio can be combined with point mutations on streptavidin to optimize the catalytic uncaging of an allyl-carbamate-protected thyroid hormone triiodothyronine. These results demonstrate that artificial metalloenzymes offer highly modular tools to perform bioorthogonal catalysis in live HEK cells., Nature Communications, 9, ISSN:2041-1723