Published papers

Identifying regulatory relationships between transcription factors (TFs) and genes is essential to understand diverse biological phenomena related to gene expression. Recently, deep learning–based models to predict TFs that bind to genes from nucleotide sequences of the target genes have been developed, yet these models are trained to predict known TFs only. Here, we developed a deep learning model, GReNIMJA (Gene Regulatory Network Inference by Mixing and Jointing features of Amino acid and nucleotide sequences), to predict gene regulation even by unknown TFs. Our model is designed to mix the features of the TF amino acid sequences and nucleotide sequences of the target genes using a 2D Long Short-Term Memory architecture and to perform binary classification with the aim of determining the presence or absence of a regulatory relationship. By explicitly modeling interactions between TFs and genes, our model can predict gene regulation for unknown TFs. The accuracy of our model in predicting regulatory relationships was 84.4% for known TFs (higher than those of conventional models) and 68.5% for unknown TFs; the latter is an unsolved task for conventional deep learning-based models. We expect our model to advance identification of unknown gene regulatory networks and contribute to the understanding of diverse biological phenomena.

Bayesian optimization is efficient even with a small amount of data and is used in engineering and in science, including biology and chemistry. In Bayesian optimization, a parameterized model with an uncertainty is fitted to explain the experimental data, and then the model suggests parameters that would most likely improve the results. Batch Bayesian optimization reduces the processing time of optimization by parallelizing experiments. However, batch Bayesian optimization cannot be applied if the number of parallelized experiments is limited by the cost or scarcity of equipment; in such cases, sequential methods require an unrealistic amount of time. In this study, we developed pipelining Bayesian optimization (PipeBO) to reduce the processing time of optimization even with a limited number of parallel experiments. PipeBO was inspired by the pipelining of central processing unit architecture, which divides computational tasks into multiple processes. PipeBO was designed to achieve experiment parallelization by overlapping various processes of the experiments. PipeBO uses the results of completed experiments to update the parameters of running parallelized experiments. PipeBO was mathematically formulated by modeling experiments as multiple processes with asynchronous result arrival, enabling partial model updates in a pipelined fashion. Using the Black-Box Optimization Benchmarking, which consists of 24 benchmark functions, we compared PipeBO with the sequential Bayesian optimization methods. PipeBO reduced the average processing time of optimization to about 56% for the experiments that consisted of two processes or even less for those with more processes for 20 out of the 24 functions. Overall, PipeBO parallelizes Bayesian optimization in experimental equipment–limited situations so that efficient optimization can be achieved.

The larva of the sleeping chironomid, Polypedilum vanderplanki, is the only insect capable of extreme desiccation tolerance, known as anhydrobiosis. The larvae can survive near-complete desiccation and, upon rehydration, rapidly resume metabolism and return to their normal life cycle. Activation of genes involved in antioxidant activity, protection of biomolecules, and DNA repair is required for desiccation tolerance. In the desiccation-tolerant P. vanderplanki cell line Pv11, the key factors of the regulatory network for these genes are heat shock factor (Hsf) and nuclear transcription factor Y subunit C (NF-YC). However, how desiccation tolerance is established at the whole-body level remains unknown. To unravel the dynamic response to desiccation at this level, it is necessary to clarify which molecules need to function for desiccation tolerance and when. Here, we newly acquired and analyzed detailed time-series gene expression data to reveal the regulatory mechanisms underlying the transition to an ametabolic state during desiccation. We showed that the acquisition of desiccation tolerance requires the expression of a number of desiccation-inducible genes in a specific order. This order can be explained by a gene regulatory network triggered by nuclear transcription factor Y subunit A (NF-YA) and Sine oculis-related homeobox 4 (Six4). To our knowledge, this is the first report that desiccation tolerance is related to stepwise changes in gene expression in response to desiccation.

The identification of gene regulatory networks is important for understanding the mechanisms of various biological phenomena. Many methods have been proposed to infer networks from time-series gene expression data obtained by high-throughput next-generation sequencings. Such methods can effectively infer gene regulatory networks for in silico data, but inferring the networks accurately from in vivo data remains a challenge because of the large noise and low time sampling rate. Here, we proposed a novel unsupervised learning method, Multi-view attention Long Short-term memory for Network inference (MaLoN). It can infer gene regulatory networks with temporal changes in gene regulation using the multi-view attention Long Short-term memory model. MaLoN inferred gene regulatory networks more accurately than 6 existing methods for in vivo benchmark Saccharomyces cerevisiae and Escherichia coli datasets with small numbers of genes, but less accurate for large-scale datasets. Therefore, MaLoN is currently effective for small-scale datasets only. The ablated models indicated that the multi-view attention mechanism suppressed false positives. The order of activation of gene regulations inferred by MaLoN was consistent with existing knowledge.

Predicting olfactory perceptions from odorant molecules is challenging due to the complex and potentially discontinuous nature of the perceptual space for smells. In this study, we introduce a deep learning model, Mol-PECO (Molecular Representation by Positional Encoding of Coulomb Matrix), designed to predict olfactory perceptions based on molecular structures and electrostatics. Mol-PECO learns the efficient embedding of molecules by utilizing the Coulomb matrix, which encodes atomic coordinates and charges, as an alternative of the adjacency matrix and its Laplacian eigenfunctions as positional encoding of atoms. With a comprehensive dataset of odor molecules and descriptors, Mol-PECO outperforms traditional machine learning methods using molecular fingerprints and graph neural networks based on adjacency matrices. The learned embeddings by Mol-PECO effectively capture the odor space, enabling global clustering of descriptors and local retrieval of similar odorants. This work contributes to a deeper understanding of the olfactory sense and its mechanisms.

Introduction: The COVID-19 Disease Map project is a large-scale community effort uniting 277 scientists from 130 Institutions around the globe. We use high-quality, mechanistic content describing SARS-CoV-2-host interactions and develop interoperable bioinformatic pipelines for novel target identification and drug repurposing. Methods: Extensive community work allowed an impressive step forward in building interfaces between Systems Biology tools and platforms. Our framework can link biomolecules from omics data analysis and computational modelling to dysregulated pathways in a cell-, tissue- or patient-specific manner. Drug repurposing using text mining and AI-assisted analysis identified potential drugs, chemicals and microRNAs that could target the identified key factors. Results: Results revealed drugs already tested for anti-COVID-19 efficacy, providing a mechanistic context for their mode of action, and drugs already in clinical trials for treating other diseases, never tested against COVID-19. Discussion: The key advance is that the proposed framework is versatile and expandable, offering a significant upgrade in the arsenal for virus-host interactions and other complex pathologies.

Pv11 is an insect cell line established from the midge Polypedilum vanderplanki, whose larval form exhibits an extreme desiccation tolerance known as anhydrobiosis. Pv11 itself is also capable of anhydrobiosis, which is induced by trehalose treatment. Here we report the successful construction of a genome editing system for Pv11 cells and its application to the identification of signaling pathways involved in anhydrobiosis. Using the Cas9-mediated gene knock-in system, we established Pv11 cells that stably expressed GCaMP3 to monitor intracellular Ca2+ mobilization. Intriguingly, trehalose treatment evoked a transient increase in cytosolic Ca2+ concentration, and further experiments revealed that the calmodulin–calcineurin–NFAT pathway contributes to tolerance of trehalose treatment as well as desiccation tolerance, while the calmodulin–calmodulin kinase–CREB pathway conferred only desiccation tolerance on Pv11 cells. Thus, our results show a critical contribution of the trehalose-induced Ca2+ surge to anhydrobiosis and demonstrate temporally different roles for each signaling pathway.

We need to effectively combine the knowledge from surging literature with complex datasets to propose mechanistic models of SARS-CoV-2 infection, improving data interpretation and predicting key targets of intervention. Here, we describe a large-scale community effort to build an open access, interoperable and computable repository of COVID-19 molecular mechanisms. The COVID-19 Disease Map (C19DMap) is a graphical, interactive representation of disease-relevant molecular mechanisms linking many knowledge sources. Notably, it is a computational resource for graph-based analyses and disease modelling. To this end, we established a framework of tools, platforms and guidelines necessary for a multifaceted community of biocurators, domain experts, bioinformaticians and computational biologists. The diagrams of the C19DMap, curated from the literature, are integrated with relevant interaction and text mining databases. We demonstrate the application of network analysis and modelling approaches by concrete examples to highlight new testable hypotheses. This framework helps to find signatures of SARS-CoV-2 predisposition, treatment response or prioritisation of drug candidates. Such an approach may help deal with new waves of COVID-19 or similar pandemics in the long-term perspective.

The Pv11, an insect cell line established from the midge Polypedilum vanderplanki, is capa-ble of extreme hypometabolic desiccation tolerance, so-called anhydrobiosis. We previously discov-ered that heat shock factor 1 (HSF1) contributes to the acquisition of desiccation tolerance by Pv11 cells, but the mechanistic details have yet to be elucidated. Here, by analyzing the gene expression profiles of newly established HSF1-knockout and-rescue cell lines, we show that HSF1 has a ge-nome-wide effect on gene regulation in Pv11. The HSF1-knockout cells exhibit a reduced desiccation survival rate, but this is completely restored in HSF1-rescue cells. By comparing mRNA profiles of the two cell lines, we reveal that HSF1 induces anhydrobiosis-related genes, especially genes encoding late embryogenesis abundant proteins and thioredoxins, but represses a group of genes involved in basal cellular processes, thus promoting an extreme hypometabolism state in the cell. In addition, HSF1 binding motifs are enriched in the promoters of anhydrobiosis-related genes and we demonstrate binding of HSF1 to these promoters by ChIP-qPCR. Thus, HSF1 directly regulates the transcription of anhydrobiosis-related genes and consequently plays a pivotal role in the induction of anhydrobiotic ability in Pv11 cells.

Copyright: © 2020 Yamada et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Water is essential for living organisms. Terrestrial organisms are incessantly exposed to the stress of losing water, desiccation stress. Avoiding the mortality caused by desiccation stress, many organisms acquired molecular mechanisms to tolerate desiccation. Larvae of the African midge, Polypedilum vanderplanki, and its embryonic cell line Pv11 tolerate desiccation stress by entering an ametabolic state, anhydrobiosis, and return to active life after rehydration. The genes related to desiccation tolerance have been comprehensively analyzed, but transcriptional regulatory mechanisms to induce these genes after desiccation or rehydration remain unclear. Here, we comprehensively analyzed the gene regulatory network in Pv11 cells and compared it with that of Drosophila melanogaster, a desiccation sensitive species. We demonstrated that nuclear transcription factor Y subunit gamma-like, which is important for drought stress tolerance in plants, and its transcriptional regulation of downstream positive feedback loops have a pivotal role in regulating various anhydrobiosis-related genes. This study provides an initial insight into the systemic mechanism of desiccation tolerance.