Machine Learning Refinement of In Situ Images Acquired by Low Electron Dose LC-TEM.

We have studied a machine learning (ML) technique for refining images acquired during in situ observation using liquid-cell transmission electron microscopy. Our model is constructed using a U-Net architecture and a ResNet encoder. For training our ML model, we prepared an original image dataset that contained pairs of images of samples acquired with and without a solution present. The former images were used as noisy images, and the latter images were used as corresponding ground truth images. The number of pairs of image sets was 1,204, and the image sets included images acquired at several different magnifications and electron doses. The trained model converted a noisy image into a clear image. The time necessary for the conversion was on the order of 10 ms, and we applied the model to in situ observations using the software Gatan DigitalMicrograph (DM). Even if a nanoparticle was not visible in a view window in the DM software because of the low electron dose, it was visible in a successive refined image generated by our ML model.

Gait video-based prediction of unified Parkinson's disease rating scale score: a retrospective study.

BACKGROUND: The diagnosis of Parkinson's disease (PD) and evaluation of its symptoms require in-person clinical examination. Remote evaluation of PD symptoms is desirable, especially during a pandemic such as the coronavirus disease 2019 pandemic. One potential method to remotely evaluate PD motor impairments is video-based analysis. In this study, we aimed to assess the feasibility of predicting the Unified Parkinson's Disease Rating Scale (UPDRS) score from gait videos using a convolutional neural network (CNN) model. METHODS: We retrospectively obtained 737 consecutive gait videos of 74 patients with PD and their corresponding neurologist-rated UPDRS scores. We utilized a CNN model for predicting the total UPDRS part III score and four subscores of axial symptoms (items 27, 28, 29, and 30), bradykinesia (items 23, 24, 25, 26, and 31), rigidity (item 22) and tremor (items 20 and 21). We trained the model on 80% of the gait videos and used 10% of the videos as a validation dataset. We evaluated the predictive performance of the trained model by comparing the model-predicted score with the neurologist-rated score for the remaining 10% of videos (test dataset). We calculated the coefficient of determination (R2) between those scores to evaluate the model's goodness of fit. RESULTS: In the test dataset, the R2 values between the model-predicted and neurologist-rated values for the total UPDRS part III score and subscores of axial symptoms, bradykinesia, rigidity, and tremor were 0.59, 0.77, 0.56, 0.46, and 0.0, respectively. The performance was relatively low for videos from patients with severe symptoms. CONCLUSIONS: Despite the low predictive performance of the model for the total UPDRS part III score, it demonstrated relatively high performance in predicting subscores of axial symptoms. The model approximately predicted the total UPDRS part III scores of patients with moderate symptoms, but the performance was low for patients with severe symptoms owing to limited data. A larger dataset is needed to improve the model's performance in clinical settings.

Fast Improvement of TEM Images with Low-Dose Electrons by Deep Learning.

Low electron dose observation is indispensable for observing various samples using a transmission electron microscope; consequently, image processing has been used to improve transmission electron microscopy (TEM) images. To apply such image processing to in situ observations, we here apply a convolutional neural network to TEM imaging. Using a dataset that includes short-exposure images and long-exposure images, we develop a pipeline for processed short-exposure images, based on end-to-end training. The quality of images acquired with a total dose of approximately $5$$e^{-}$ per pixel becomes comparable to that of images acquired with a total dose of approximately $1{,}000$$e^{-}$ per pixel. Because the conversion time is approximately 8 ms, in situ observation at 125 fps is possible. This imaging technique enables in situ observation of electron-beam-sensitive specimens.

Dual graph convolutional neural network for predicting chemical networks.

BACKGROUND: Predicting of chemical compounds is one of the fundamental tasks in bioinformatics and chemoinformatics, because it contributes to various applications in metabolic engineering and drug discovery. The recent rapid growth of the amount of available data has enabled applications of computational approaches such as statistical modeling and machine learning method. Both a set of chemical interactions and chemical compound structures are represented as graphs, and various graph-based approaches including graph convolutional neural networks have been successfully applied to chemical network prediction. However, there was no efficient method that can consider the two different types of graphs in an end-to-end manner. RESULTS: We give a new formulation of the chemical network prediction problem as a link prediction problem in a graph of graphs (GoG) which can represent the hierarchical structure consisting of compound graphs and an inter-compound graph. We propose a new graph convolutional neural network architecture called dual graph convolutional network that learns compound representations from both the compound graphs and the inter-compound network in an end-to-end manner. CONCLUSIONS: Experiments using four chemical networks with different sparsity levels and degree distributions shows that our dual graph convolution approach achieves high prediction performance in relatively dense networks, while the performance becomes inferior on extremely-sparse networks.

Exploring phenotype patterns of breast cancer within somatic mutations: a modicum in the intrinsic code.

Triple-negative (TN) breast cancer (BC) patients have limited treatment options and poor prognosis even after extant treatments and standard chemotherapeutic regimens. Linking TN patients to clinically known phenotypes with appropriate treatments is vital. Location-specific sequence variants are expected to be useful for this purpose by identifying subgroups within a disease population. Single gene mutational signatures have been widely reported, with related phenotypes in literature. We thoroughly survey currently available mutations (and mutated genes), linked to BC phenotypes, to demonstrate their limited performance as sole predictors/biomarkers to assign phenotypes to patients. We then explore mutational combinations, as a pilot study, using The Cancer Genome Atlas Research Network mutational data of BC and three machine learning methods: association rules (limitless arity multiple procedure), decision tree and hierarchical disjoint clustering. The study results in a patient classification scheme through combinatorial mutations in Phosphatidylinositol-4,5-Bisphosphate 3-Kinase and tumor protein 53, being consistent with all three methods, implying its validity from a diverse viewpoint. However, it would warrant further research to select multi-gene signatures to identify phenotypes specifically and be clinically used routinely.

Predictions of Cleavability of Calpain Proteolysis by Quantitative Structure-Activity Relationship Analysis Using Newly Determined Cleavage Sites and Catalytic Efficiencies of an Oligopeptide Array.

Calpains are intracellular Ca(2+)-regulated cysteine proteases that are essential for various cellular functions. Mammalian conventional calpains (calpain-1 and calpain-2) modulate the structure and function of their substrates by limited proteolysis. Thus, it is critically important to determine the site(s) in proteins at which calpains cleave. However, the calpains' substrate specificity remains unclear, because the amino acid (aa) sequences around their cleavage sites are very diverse. To clarify calpains' substrate specificities, 84 20-mer oligopeptides, corresponding to P10-P10' of reported cleavage site sequences, were proteolyzed by calpains, and the catalytic efficiencies (kcat/Km) were globally determined by LC/MS. This analysis revealed 483 cleavage site sequences, including 360 novel ones. Thekcat/Kms for 119 sites ranged from 12.5-1,710 M(-1)s(-1) Although most sites were cleaved by both calpain-1 and -2 with a similarkcat/Km, sequence comparisons revealed distinct aa preferences at P9-P7/P2/P5'. The aa compositions of the novel sites were not statistically different from those of previously reported sites as a whole, suggesting calpains have a strict implicit rule for sequence specificity, and that the limited proteolysis of intact substrates is because of substrates' higher-order structures. Cleavage position frequencies indicated that longer sequences N-terminal to the cleavage site (P-sites) were preferred for proteolysis over C-terminal (P'-sites). Quantitative structure-activity relationship (QSAR) analyses using partial least-squares regression and >1,300 aa descriptors achievedkcat/Kmprediction withr= 0.834, and binary-QSAR modeling attained an 87.5% positive prediction value for 132 reported calpain cleavage sites independent of our model construction. These results outperformed previous calpain cleavage predictors, and revealed the importance of the P2, P3', and P4' sites, and P1-P2 cooperativity. Furthermore, using our binary-QSAR model, novel cleavage sites in myoglobin were identified, verifying our predictor. This study increases our understanding of calpain substrate specificities, and opens calpains to "next-generation,"i.e.activity-related quantitative and cooperativity-dependent analyses.

Machine learning reveals orbital interaction in materials.

We propose a novel representation of materials named an 'orbital-field matrix (OFM)', which is based on the distribution of valence shell electrons. We demonstrate that this new representation can be highly useful in mining material data. Experimental investigation shows that the formation energies of crystalline materials, atomization energies of molecular materials, and local magnetic moments of the constituent atoms in bimetal alloys of lanthanide metal and transition-metal can be predicted with high accuracy using the OFM. Knowledge regarding the role of the coordination numbers of the transition-metal and lanthanide elements in determining the local magnetic moments of the transition-metal sites can be acquired directly from decision tree regression analyses using the OFM.

Similarity-based machine learning methods for predicting drug-target interactions: a brief review.

Computationally predicting drug-target interactions is useful to select possible drug (or target) candidates for further biochemical verification. We focus on machine learning-based approaches, particularly similarity-based methods that use drug and target similarities, which show relationships among drugs and those among targets, respectively. These two similarities represent two emerging concepts, the chemical space and the genomic space. Typically, the methods combine these two types of similarities to generate models for predicting new drug-target interactions. This process is also closely related to a lot of work in pharmacogenomics or chemical biology that attempt to understand the relationships between the chemical and genomic spaces. This background makes the similarity-based approaches attractive and promising. This article reviews the similarity-based machine learning methods for predicting drug-target interactions, which are state-of-the-art and have aroused great interest in bioinformatics. We describe each of these methods briefly, and empirically compare these methods under a uniform experimental setting to explore their advantages and limitations.

Ribosomes in a stacked array: elucidation of the step in translation elongation at which they are stalled during S-adenosyl-L-methionine-induced translation arrest of CGS1 mRNA.

Expression of CGS1, which codes for an enzyme of methionine biosynthesis, is feedback-regulated by mRNA degradation in response to S-adenosyl-L-methionine (AdoMet). In vitro studies revealed that AdoMet induces translation arrest at Ser-94, upon which several ribosomes stack behind the arrested one, and mRNA degradation occurs at multiple sites that presumably correspond to individual ribosomes in a stacked array. Despite the significant contribution of stacked ribosomes to inducing mRNA degradation, little is known about the ribosomes in the stacked array. Here, we assigned the peptidyl-tRNA species of the stacked second and third ribosomes to their respective codons and showed that they are arranged at nine-codon intervals behind the Ser-94 codon, indicating tight stacking. Puromycin reacts with peptidyl-tRNA in the P-site, releasing the nascent peptide as peptidyl-puromycin. This reaction is used to monitor the activity of the peptidyltransferase center (PTC) in arrested ribosomes. Puromycin reaction of peptidyl-tRNA on the AdoMet-arrested ribosome, which is stalled at the pre-translocation step, was slow. This limited reactivity can be attributed to the peptidyl-tRNA occupying the A-site at this step rather than to suppression of PTC activity. In contrast, puromycin reactions of peptidyl-tRNA with the stacked second and third ribosomes were slow but were not as slow as pre-translocation step ribosomes. We propose that the anticodon end of peptidyl-tRNA resides in the A-site of the stacked ribosomes and that the stacked ribosomes are stalled at an early step of translocation, possibly at the P/E hybrid state.

Graph Network-Based Simulation of Multicellular Dynamics Driven by Concentrated Polymer Brush-Modified Cellulose Nanofibers.

Manipulating the three-dimensional (3D) structures of cells is important for facilitating to repair or regenerate tissues. A self-assembly system of cells with cellulose nanofibers (CNFs) and concentrated polymer brushes (CPBs) has been developed to fabricate various cell 3D structures. To further generate tissues at an implantable level, it is necessary to carry out a large number of experiments using different cell culture conditions and material properties; however this is practically intractable. To address this issue, we present a graph-neural network-based simulator (GNS) that can be trained by using assembly process images to predict the assembly status of future time steps. A total of 24 (25 steps) time-series images were recorded (four repeats for each of six different conditions), and each image was transformed into a graph by regarding the cells as nodes and the connecting neighboring cells as edges. Using the obtained data, the performances of the GNS were examined under three scenarios (i.e., changing a pair of the training and testing data) to verify the possibility of using the GNS as a predictor for further time steps. It was confirmed that the GNS could reasonably reproduce the assembly process, even under the toughest scenario, in which the experimental conditions differed between the training and testing data. Practically, this means that the GNS trained by the first 24 h images could predict the cell types obtained 3 weeks later. This result could reduce the number of experiments required to find the optimal conditions for generating cells with desired 3D structures. Ultimately, our approach could accelerate progress in regenerative medicine.

ROS-DET: robust detector of switching mechanisms in gene expression.

A switching mechanism in gene expression, where two genes are positively correlated in one condition and negatively correlated in the other condition, is a key to elucidating complex biological systems. There already exist methods for detecting switching mechanisms from microarrays. However, current approaches have problems under three real cases: outliers, expression values with a very small range and a small number of examples. ROS-DET overcomes these three problems, keeping the computational complexity of current approaches. We demonstrated that ROS-DET outperformed existing methods, under that all these three situations are considered. Furthermore, for each of the top 10 pairs ranked by ROS-DET, we attempted to identify a pathway, i.e. consecutive biological phenomena, being related with the corresponding two genes by checking the biological literature. In 8 out of the 10 pairs, we found two parallel pathways, one of the two genes being in each of the two pathways and two pathways coming to (or starting with) the same gene. This indicates that two parallel pathways would be cooperatively used under one experimental condition, corresponding to the positive correlation, and the two pathways might be alternatively used under the other condition, corresponding to the negative correlation. ROS-DET is available from http://www.bic.kyoto-u.ac.jp/pathway/kayano/ros-det.htm.

Accelerated discovery of multi-elemental reverse water-gas shift catalysts using extrapolative machine learning approach.

Designing novel catalysts is key to solving many energy and environmental challenges. Despite the promise that data science approaches, including machine learning (ML), can accelerate the development of catalysts, truly novel catalysts have rarely been discovered through ML approaches because of one of its most common limitations and criticisms-the assumed inability to extrapolate and identify extraordinary materials. Herein, we demonstrate an extrapolative ML approach to develop new multi-elemental reverse water-gas shift catalysts. Using 45 catalysts as the initial data points and performing 44 cycles of the closed loop discovery system (ML prediction + experiment), we experimentally tested a total of 300 catalysts and identified more than 100 catalysts with superior activity compared to those of the previously reported high-performance catalysts. The composition of the optimal catalyst discovered was Pt(3)/Rb(1)-Ba(1)-Mo(0.6)-Nb(0.2)/TiO2. Notably, niobium (Nb) was not included in the original dataset, and the catalyst composition identified was not predictable even by human experts.

Early Detection of Nucleation Events From Solution in LC-TEM by Machine Learning.

To support the detection, recording, and analysis of nucleation events during in situ observations, we developed an early detection system for nucleation events observed using a liquid-cell transmission electron microscope. Detectability was achieved using the machine learning equivalent of detection by humans watching a video numerous times. The detection system was applied to the nucleation of sodium chloride crystals from a saturated acetone solution of sodium chlorate. Nanoparticles with a radius of more greater than 150 nm were detected in a viewing area of 12 µm × 12 µm by the detection system. The analysis of the change in the size of the growing particles as a function of time suggested that the crystal phase of the particles with a radius smaller than 400 nm differed from that of the crystals larger than 400 nm. Moreover, the use of machine learning enabled the detection of numerous nanometer sized nuclei. The nucleation rate estimated from the machine-learning-based detection was of the same order as that estimated from the detection using manual procedures.

Calcium sparks enhance the tissue fluidity within epithelial layers and promote apical extrusion of transformed cells.

In vertebrates, newly emerging transformed cells are often apically extruded from epithelial layers through cell competition with surrounding normal epithelial cells. However, the underlying molecular mechanism remains elusive. Here, using phospho-SILAC screening, we show that phosphorylation of AHNAK2 is elevated in normal cells neighboring RasV12 cells soon after the induction of RasV12 expression, which is mediated by calcium-dependent protein kinase C. In addition, transient upsurges of intracellular calcium, which we call calcium sparks, frequently occur in normal cells neighboring RasV12 cells, which are mediated by mechanosensitive calcium channel TRPC1 upon membrane stretching. Calcium sparks then enhance cell movements of both normal and RasV12 cells through phosphorylation of AHNAK2 and promote apical extrusion. Moreover, comparable calcium sparks positively regulate apical extrusion of RasV12-transformed cells in zebrafish larvae as well. Hence, calcium sparks play a crucial role in the elimination of transformed cells at the early phase of cell competition.

Mining metabolic pathways through gene expression.

MOTIVATION: An observed metabolic response is the result of the coordinated activation and interaction between multiple genetic pathways. However, the complex structure of metabolism has meant that a compete understanding of which pathways are required to produce an observed metabolic response is not fully understood. In this article, we propose an approach that can identify the genetic pathways which dictate the response of metabolic network to specific experimental conditions. RESULTS: Our approach is a combination of probabilistic models for pathway ranking, clustering and classification. First, we use a non-parametric pathway extraction method to identify the most highly correlated paths through the metabolic network. We then extract the defining structure within these top-ranked pathways using both Markov clustering and classification algorithms. Furthermore, we define detailed node and edge annotations, which enable us to track each pathway, not only with respect to its genetic dependencies, but also allow for an analysis of the interacting reactions, compounds and KEGG sub-networks. We show that our approach identifies biologically meaningful pathways within two microarray expression datasets using entire KEGG metabolic networks. AVAILABILITY AND IMPLEMENTATION: An R package containing a full implementation of our proposed method is currently available from http://www.bic.kyoto-u.ac.jp/pathway/timhancock.

Efficiently finding genome-wide three-way gene interactions from transcript- and genotype-data.

MOTIVATION: We address the issue of finding a three-way gene interaction, i.e. two interacting genes in expression under the genotypes of another gene, given a dataset in which expressions and genotypes are measured at once for each individual. This issue can be a general, switching mechanism in expression of two genes, being controlled by categories of another gene, and finding this type of interaction can be a key to elucidating complex biological systems. The most suitable method for this issue is likelihood ratio test using logistic regressions, which we call interaction test, but a serious problem of this test is computational intractability at a genome-wide level. RESULTS: We developed a fast method for this issue which improves the speed of interaction test by around 10 times for any size of datasets, keeping highly interacting genes with an accuracy of approximately 85%. We applied our method to approximately 3 x 10(8) three-way combinations generated from a dataset on human brain samples and detected three-way gene interactions with small P-values. To check the reliability of our results, we first conducted permutations by which we can show that the obtained P-values are significantly smaller than those obtained from permuted null examples. We then used GEO (Gene Expression Omnibus) to generate gene expression datasets with binary classes to confirm the detected three-way interactions by using these datasets and interaction tests. The result showed us some datasets with significantly small P-values, strongly supporting the reliability of the detected three-way interactions. AVAILABILITY: Software is available from http://www.bic.kyoto-u.ac.jp/pathway/kayano/bioinfo_three-way.html CONTACT: kayano@kuicr.kyoto-u.ac.jp SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

On the performance of methods for finding a switching mechanism in gene expression.

We address an issue of detecting a switching mechanism in gene expression, where two genes are positively correlated for one experimental condition while they are negatively correlated for another. We compare the performance of existing methods for this issue, roughly divided into two types: interaction test (IT) and the difference of correlation coefficients. Interaction test, currently a standard approach for detecting epistasis in genetics, is the log-likelihood ratio test between two logistic regressions with/without an interaction term, resulting in checking the strength of interaction between two genes. On the other hand, two correlation coefficients can be computed for two experimental conditions and the difference of them shows the alteration of expression trends in a more straightforward manner. In our experiments, we tested three different types of correlation coefficients: Pearson, Spearman and a midcorrelation (biweight midcorrelation). The experiment was performed by using ~ 2.3 × 10(9) combinations selected out of the GEO (Gene Expression Omnibus) database. We sorted all combinations according to the p-values of IT or by the absolute values of the difference of correlation coefficients and then visually evaluated the top ranked combinations in terms of the switching mechanism. The result showed that 1) combinations detected by IT included non-switching combinations and 2) Pearson was affected by outliers easily while Spearman and the midcorrelation seemed likely to avoid them.

CaMPDB: a resource for calpain and modulatory proteolysis.

While the importance of modulatory proteolysis in research has steadily increased, knowledge on this process has remained largely disorganized, with the nature and role of entities composing modulatory proteolysis still uncertain. We built CaMPDB, a resource on modulatory proteolysis, with a focus on calpain, a well-studied intracellular protease which regulates substrate functions by proteolytic processing. CaMPDB contains sequences of calpains, substrates and inhibitors as well as substrate cleavage sites, collected from the literature. Some cleavage efficiencies were evaluated by biochemical experiments and a cleavage site prediction tool is provided to assist biologists in understanding calpain-mediated cellular processes. CaMPDB is freely accessible at http://calpain.org.

The role of Mediator and Little Elongation Complex in transcription termination.

Mediator is a coregulatory complex that regulates transcription of Pol II-dependent genes. Previously, we showed that human Mediator subunit MED26 plays a role in the recruitment of Super Elongation Complex (SEC) or Little Elongation Complex (LEC) to regulate the expression of certain genes. MED26 plays a role in recruiting SEC to protein-coding genes including c-myc and LEC to small nuclear RNA (snRNA) genes. However, how MED26 engages SEC or LEC to regulate distinct genes is unclear. Here, we provide evidence that MED26 recruits LEC to modulate transcription termination of non-polyadenylated transcripts including snRNAs and mRNAs encoding replication-dependent histone (RDH) at Cajal bodies. Our findings indicate that LEC recruited by MED26 promotes efficient transcription termination by Pol II through interaction with CBC-ARS2 and NELF/DSIF, and promotes 3' end processing by enhancing recruitment of Integrator or Heat Labile Factor to snRNA or RDH genes, respectively.

Probabilistic path ranking based on adjacent pairwise coexpression for metabolic transcripts analysis.

MOTIVATION: Pathway knowledge in public databases enables us to examine how individual metabolites are connected via chemical reactions and what genes are implicated in those processes. For two given (sets of) compounds, the number of possible paths between them in a metabolic network can be intractably large. It would be informative to rank these paths in order to differentiate between them. RESULTS: Focusing on adjacent pairwise coexpression, we developed an algorithm which, for a specified k, efficiently outputs the top k paths based on a probabilistic scoring mechanism, using a given metabolic network and microarray datasets. Our idea of using adjacent pairwise coexpression is supported by recent studies that local coregulation is predominant in metabolism. We first evaluated this idea by examining to what extent highly correlated gene pairs are adjacent and how often they are consecutive in a metabolic network. We then applied our algorithm to two examples of path ranking: the paths from glucose to pyruvate in the entire metabolic network of yeast and the paths from phenylalanine to sinapyl alcohol in monolignols pathways of arabidopsis under several different microarray conditions, to confirm and discuss the performance analysis of our method.