Biology subjects

Fujii, S.

Publications and source records attributed to Fujii, S..

At least 37 records · Page 2Linked to original sources

bioRxiv2024

The KU70-SAP domain has an overlapping function with DNA-PKcs in limiting the lateral movement of KU along DNA

The non-homologous end-joining (NHEJ) pathway is critical for DNA double-strand break repair and is essential for lymphocyte development and maturation. The Ku70/Ku80 heterodimer (KU) binds to DNA ends, initiating NHEJ and recruiting additional factors, including DNA-dependent protein kinase catalytic subunit (DNA-PKcs) that caps the ends and pushes KU inward. The C-terminus of Ku70 in higher eukaryotes includes a flexible linker and a SAP domain, whose physiological role remains poorly understood. To investigate this, we generated a mouse model with knock-in deletion of the SAP domain (Ku70{Delta}SAP/{Delta}SAP). Ku70{Delta}SAP supports KU stability and its recruitment to DNA damage sites in vivo. In contrast to the growth retardation and immunodeficiency seen in Ku70-/- mice, Ku70{Delta}SAP/{Delta}SAP mice show no defects in lymphocyte development and maturation. Structural modeling of KU on long dsDNA, but not dsRNA suggests that the SAP domain can bind to an adjacent major groove, where it can limit KUs rotation and lateral movement along the dsDNA. Accordingly, in the absence of DNA-PKcs that caps the ends, Ku70{Delta}SAP fails to support stable DNA damage-induced KU foci. In DNA-PKcs-/- mice, Ku70{Delta}SAP abrogates the leaky T cell development and reduces both the qualitative and quantitative aspects of residual V(D)J recombination. In the absence of DNA-PKcs, purified Ku70{Delta}SAP has reduced affinity for DNA ends and dissociates more readily at lower concentration and accumulated as multimers at high concentration. These findings revealed a physiological role of the SAP domain in NHEJ by restricting KU rotation and lateral movement on DNA that is largely masked by DNA-PKcs. HighlightKu70 is a conserved non-homologous end-joining (NHEJ) factor. Using genetically engineered mouse models and biochemical analyses, our study uncovered a previously unappreciated role of the C-terminal SAP domain of Ku70 in limiting the lateral movement of KU on DNA ends and ensuring end protection. The presence of DNA-PKcs partially masks this role of the SAP domain.

genetics↗

bioRxiv2024

Oscillations create groove: A reservoir model for learning complex drumming performances

Musical performances, particularly in drumming, are characterized not only by their structured rhythmic patterns but also by the subtle variations in timing and amplitude series that create expressive complexity. This study proposes a neural-inspired computational model to investigate how the brain might learn and internalizes such complex rhythms. Inspired by the established roles of the cerebellum and basal ganglia in production of rhythms and timings, we utilize an oscillation-driven reservoir computer, a recurrent neural network model for temporal learning, to simulate the generation of human-like expressive drumming performances. First, the model was trained to replicate Jeff Porcaros distinctive hi-hat patterns. Analyses revealed that the outputs of the model incorporating high-frequency oscillators ([50, 100] Hz), closely matched the original drumming, reproducing its characteristic fluctuations and patterns in inter-beat timings (microtiming) and amplitudes. Next, the model was trained to generate multidimensional drum kit performances for various genres (funk, jazz, samba, and rock). The models outputs exhibited timing deviation and audio features characteristic of the original performances. Our findings demonstrate that oscillation-driven reservoir computing can replicate the rhythmic complexity of professional drumming, suggesting it as a potential computational principle for motor timing and rhythm generation. This approach provides a powerful framework for understanding how the brain generates and processes intricate rhythmic patterns.

neuroscience↗

bioRxiv2024

Drumming Performance and Underlying Muscle Activities in a Professional Rock Drummer with Lower-Limb Dystonia: A Case Study

Task-specific focal dystonia (TSFD), characterized by the loss of fine motor control and coordination, affects drummers lower-limb movements, yet it remains relatively understudied, leaving a gap in understanding its effects on drumming performance and underlying muscle activities. This study explores lower limb dystonias impact on drumming performance and underlying muscle activity in a professional rock drummer. The drummer executed an eight-beat pattern on a drum kit, emphasizing kicking the bass drum on the initial beat and performing syncopation of the third beat. Dystonia symptoms primarily manifested in the initial beat, with fewer symptoms on syncopation of the third beat. Analysis revealed decreased bass-drum sound peak amplitude during the initial beat and increased (becoming more negative) synchronization error. Electromyographic (EMG) measurements of ten muscles in the affected right lower limb showed significant changes in the Biceps Femoris (BF), Tibialis Anterior (TA), Extensor Digitorum Longus (EDL), and Extensor Digitorum Brevis (EDB) muscles during symptom onset. Notably, we observed 1) earlier overactivation of the TA and EDL muscles during the leg lift-up motion or preparatory phase of pedaling, 2) reduced activation of the EDB muscle, and 3) increased activation of the BF muscle during the final pedaling movement when symptoms occurred. These findings suggest that lower-limb dystonia symptoms are characterized by a reduction in amplitude and an increase in synchronization error, potentially due to premature overactivation of the ankle dorsiflexor muscles.

neuroscience↗

bioRxiv2024

Long-Term Monolayer Cultivation Captures Homeostatic and Regenerative Features of Human Colonic Epithelial Cells

Primary intestinal epithelial stem cell culture methods have significantly advanced understanding of mammalian intestinal development and disease. However, progress has been hampered by inconsistent methodological reporting and challenges comparing in vitro systems with in vivo observations. We previously established a unique method for long-term 2{-}dimensional (2D) cultivation of mouse lECs using an air-liquid interface (ALI) technique which appears to faithfully recapitulate homeostatic and regenerative features in vitro. Here, we further refined these methods and optimized protocols for long-term, self-organizing 2D cultivation of human lECs. During the culture, we observed that epithelial cells undergo a dynamic morphological transition from squamous to columnar shape. Using single cell transcriptomics, we identified both major lineages and minor populations, including enteroendocrine cells, tuft cells, and BEST4/CA7+ cells. Leveraging the power and scalability of a biomedical foundation model (BMFM) trained on single cell RNA sequencing data, we performed classification tasks to identify cell types across sample sources and to quantitatively benchmark our in vitro differentiated cells against cells collected from patient biopsies. We observed a striking degree of similarity between our in vitro differentiated cells and the corresponding cell types in vivo for multiple differentiated lineages. This novel approach using BMFM holds promise to expand our understanding of the regulatory mechanisms including gene-gene regulation underlying homeostasis and regeneration as well as the functions of rare and poorly understood lineages within the human intestinal epithelia. Moreover, these methods are generalizable to other organs and can be used to assess the correspondence of cells across experimental modalities. SignificanceThis manuscript addresses challenges in quantitative comparisons of cell types grown in vitro vs their in vivo counterparts. Here we use the intestinal epithelium as a model system to address this challenge. We devised an in vitro culture platform that supports multipotent intestinal epithelial stem cells and their numerous differentiated progeny. This system gives rise to all known rare and abundant lineages in correct proportions. Novel use of biomedical foundation models pre-trained on publicly available data and then fine-tuned to data from this platform enabled demonstration of high concordance of multiple in vitro differentiated lineages with the corresponding cells in vivo.

cell biology↗

bioRxiv2023

Influence of Monaural Auditory Stimulation Combined with Music on Brain Activity

Recently, the increasing attention to mental states and psychophysical health has fueled the research into methods that can aid in relaxation and recovery. Traditional methods like meditation and sauna, while effective, have their limitations; thus, the need for more accessible and convenient alternatives. Our innovative approach combines monaural beats with music, attempting to replicate the relaxing effects of a sauna in the auditory domain. In comparison to normal music and silent condition, the power of the theta active band significantly increased when listening to our modified music. Furthermore, after listening to modified music, there was a significant increase in mismatch negativity (MMN) amplitude in the oddball task. Additionally, participants subjective responses to a questionnaire indicated significant changes in body relaxation and other metrics after listening to the processed music. This state is considered similar to the "totonou" state previously reported for the sauna. Thus, the present research proposes a convenient method for achieving relaxation, opening an avenue for individuals to customize their "totonou" music based on personal preferences.

neuroscience↗

bioRxiv2023

Gamma Music: A New Acoustic Stimulus for Gamma-frequency Auditory Steady-State Response

A frequency range exceeding approximately 30 Hz, denoted as the gamma frequency range, is associated with various cognitive functions, consciousness, sensory integration, short-term memory, working memory, encoding and maintenance of episodic memory, and retrieval processes. In this study, we proposed a new form of gamma stimulation, called gamma music, combining 40 Hz auditory stimuli and music. This gamma music consists of drums, bass, and keyboard sounds, each containing a 40 Hz frequency oscillation. Since 40 Hz stimuli are known to induce an auditory steady-state response (ASSR), we used the 40 Hz power and phase locking index (PLI) as indices of neural activity during sound stimulation. We also recorded subjective ratings of each sound through a questionnaire using a visual analog scale. The gamma music, gamma drums, gamma bass, and gamma keyboard sounds showed significantly higher values in 40 Hz power and PLI compared to the control music without a 40 Hz oscillation. Particularly, the gamma keyboard sound showed a potential to induce strong ASSR, showing high values in these indices. In the subjective ratings, the gamma music, especially the gamma keyboard sound, received more relaxed, comfortable, preferred, pleasant, and natural impressions compared to the control music with conventional gamma stimulation. These results indicate that our proposed gamma music has potential as a new method for inducing ASSR. Particularly, the gamma keyboard sound proved to be an effective acoustic source for inducing a strong ASSR while preserving the comfortable and pleasant sensation of listening to music.

neuroscience↗

bioRxiv2023

Membrane contact sites regulate vacuolar fission via sphingolipid metabolism

Membrane contact sites (MCSs) are junctures that perform important roles including coordinating lipid metabolism. Previous studies have indicated that vacuolar fission/fusion processes are coupled with modifications in the membrane lipid composition. However, it has been still unclear whether MCS-mediated lipid metabolism controls the vacuolar morphology. Here we report that deletion of tricalbins (Tcb1, Tcb2, Tcb3), tethering proteins at endoplasmic reticulum (ER)-plasma membrane (PM) and ER-Golgi contact sites, alters fusion/fission dynamics and causes vacuolar fragmentation in the yeast Saccharomyces cerevisiae. In addition, we show that the sphingolipid precursor phytosphingosine accumulates in tricalbin-deleted cells, triggering the vacuolar division. Detachment of the nucleus vacuole junction (NVJ), an important contact site between the vacuole and the perinuclear ER, restored vacuolar morphology in both cells subjected to high exogenous phytosphingosine and Tcb3-deleted cells, supporting that phytosphingosine transport across the NVJ induces vacuole division. Thus, our results suggest that vacuolar morphology is maintained by MCSs through the metabolism of sphingolipids.

cell biology↗

bioRxiv2023

Plastid anionic lipids are required for membrane development and protochlorophyllide synthesis in etioplasts

Dark-germinated angiosperms develop the chloroplast precursors called etioplasts in cotyledon cells. Etioplasts develop lattice membrane structures called prolamellar bodies (PLBs), where the chlorophyll intermediate protochlorophyllide (Pchlide) forms a ternary complex with NADPH and light-dependent NADPH-Pchlide oxidoreductase (LPOR). The lipid bilayers of etioplast membranes are mainly composed of galactolipids, which play important roles in membrane-associated processes in etioplasts. Although etioplast membranes also contain two anionic lipids, phosphatidylglycerol (PG) and sulfoquinovosyldiacylglycerol (SQDG), the roles of these anionic lipids are unknown. To reveal the importance of PG and SQDG for the development of etioplasts, we characterized etiolated Arabidopsis mutants deficient in the biosynthesis of PG and SQDG. A partial deficiency in PG biosynthesis loosened the lattice structure of PLBs and impaired the insertion of Mg2+ into protoporphyrin IX, leading to a significant decrease in Pchlide content. Although a complete lack of SQDG biosynthesis did not notably affect both PLB formation and Pchlide biosynthesis, the lack of SQDG in addition to the partial deficiency of PG caused strong impairments of these processes. The results suggested that PG is required for PLB formation and Pchlide biosynthesis, whereas SQDG plays an auxiliary role in these processes. Notably, the PG deficiency and the lack of SQDG oppositely affected the dynamics of LPOR complexes after photoconversion, suggesting different involvements of PG and SQDG in the organization of LPOR complexes. Our data demonstrate pleiotropic roles of anionic lipids in etioplast development.

plant biology↗

bioRxiv2023

Transcription-directed membrane association organizes the chloroplast nucleoid structure

DNA is organized into chromatin-like structures, which support the maintenance and regulation of genomes. A unique and poorly understood form of DNA packaging exists in chloroplasts, which are endosymbiotic organelles responsible for photosynthesis. Chloroplast genomes, together with associated proteins, form membraneless structures known as nucleoids. The internal arrangement of the nucleoid, molecular mechanisms of DNA packaging, and connections between nucleoid structure and gene expression remain mostly unknown. We show that Arabidopsis thaliana chloroplast nucleoids have a unique organization driven by DNA binding to the thylakoid membranes. DNA associated with the membranes has high protein occupancy, reduced DNA accessibility, and is highly transcribed. In contrast, genes with low levels of transcription are further away from the membranes, have lower protein occupancy, and higher DNA accessibility. Disruption of transcription at specific genes in sigma factor mutants causes a corresponding reduction in membrane association, indicating that RNA polymerase activity causes DNA tethering to the membranes. We propose that transcription organizes the chloroplast nucleoid into a transcriptionally active membrane-associated core and a less active periphery.

plant biology↗

bioRxiv2023

In depth amino acid mutational analysis of the key interspecific incompatibility transmembrane factor Stigmatic Privacy 1

O_LIIn plants, there is an active prezygotic interspecific-incompatibility mechanism to prevent unfavorable hybrids between two species. We previously reported that an uncharacterized protein with four-transmembrane domains, named as Stigmatic Privacy 1 (SPRI1), is responsible for rejecting hetero-specific pollen grains in Arabidopsis thaliana. C_LIO_LIWe have conducted a functional study of the SPRI1 protein, via point-mutational experiments and biochemical analysis. We studied the molecular regulatory mechanisms of SPRI1 and the relationships with its function. C_LIO_LIThe alanine- and glycine-scanning experiments together with the evolutional analysis showed that the structural integrity of the C-terminal regions of the extracellular domain of this protein is important for its function. In addition, we found two cysteines (C67 and C80) within the extracellular domain that may be involved in the formation of intermolecular disulfide bonds. SPRI1 may form homo-multimers and is present as part of a ca. 300 kDa complex. C_LIO_LIOur present study indicates that molecular complex formation ability of SPRI1 may be important to maintain its stability and interspecific incompatibility functions in cells. C_LI

plant biology↗

bioRxiv2022

A land plant specific VPS13 mediates polarized vesicle trafficking in germinating pollen

Pollen has an extraordinary ability to convert from a dry state to an extremely rapidly growing state. During pollination, pollen receives water and Ca2+ from the contacting pistil, which will be a directional cue for pollen tube germination. The subsequent rapid activation of directional vesicular transport must support the pollen tube growth, but the molecular mechanism leading to this process is largely unknown. Here we show that a plant-specific VPS13, AtVPS13a, mediates vesicle trafficking during the polarization process in Arabidopsis pollen. AtVPS13a knockout severely affected pollen germination and lipid droplet discharge, while Ca2+-dynamics after pollination was unchanged. Cellular distribution patterns of AtVPS13a and a secretory vesicle marker were synchronized, with a slight delay to the Ca2+-dynamics in polarizing pollen. The absence of AtVPS13a led to reduced cell wall deposition during pollen germination. These results suggest that AtVPS13a mediates pollen polarization, by regulating proper directional vesicular transport following Ca2+-signaling.

plant biology↗

bioRxiv2022

High-resolution map of plastid encoded polymerase binding patterns demonstrates a major role of transcription in chloroplast gene expression

Plastids are endosymbiotic organelles containing their own genomes, which are transcribed by two types of RNA polymerases. One of those enzymes is a bacterial-type, multi-subunit polymerase encoded by the plastid genome. The plastid encoded RNA polymerase (PEP) is required for efficient expression of genes encoding proteins involved in photosynthesis. Despite the importance of PEP, its DNA binding locations have not been studied on the genome-wide scale at high resolution. We established a highly specific approach to detect the genome-wide pattern of PEP binding to chloroplasts DNA using ptChIP-seq. We found that in mature Arabidopsis thaliana chloroplasts, PEP has a complex DNA binding pattern with preferential association at genes encoding rRNA, tRNA, and a subset of photosynthetic proteins. Sigma factors SIG2 and SIG6 strongly impact PEP binding to a subset of tRNA genes and have more moderate effects on PEP binding throughout the rest of the genome. PEP binding is commonly enriched on gene promoters, around transcription start sites. Finally, the levels of PEP binding to DNA are correlated with the levels of RNA accumulation, which allowed estimating the quantitative contribution of transcription to RNA accumulation.

plant biology↗

bioRxiv2022

Global monitoring of soil animal communities using a common methodology

Here we introduce the Soil BON Foodweb Team, a cross-continental collaborative network that aims to monitor soil animal communities and food webs using consistent methodology at a global scale. Soil animals support vital soil processes via soil structure modification, direct consumption of dead organic matter, and interactions with microbial and plant communities. Soil animal effects on ecosystem functions have been demonstrated by correlative analyses as well as in laboratory and field experiments, but these studies typically focus on selected animal groups or species at one or few sites with limited variation in environmental conditions. The lack of comprehensive harmonised large-scale soil animal community data including microfauna, mesofauna, and macrofauna, in conjunction with related soil functions, limits our understanding of biological interactions in soil communities and how these interactions affect ecosystem functioning. To provide such data, the Soil BON Foodweb Team invites researchers worldwide to use a common methodology to address six long-term goals: (1) to collect globally representative harmonised data on soil micro-, meso-, and macrofauna communities; (2) to describe key environmental drivers of soil animal communities and food webs; (3) to assess the efficiency of conservation approaches for the protection of soil animal communities; (4) to describe soil food webs and their association with soil functioning globally; (5) to establish a global research network for soil biodiversity monitoring and collaborative projects in related topics; (6) to reinforce local collaboration networks and expertise and support capacity building for soil animal research around the world. In this paper, we describe the vision of the global research network and the common sampling protocol to assess soil animal communities and advocate for the use of standard methodologies across observational and experimental soil animal studies. We will use this protocol to conduct soil animal assessments and reconstruct soil food webs on the sites included in the global soil biodiversity monitoring network, Soil BON, allowing us to assess linkages among soil biodiversity, vegetation, soil physico-chemical properties, and ecosystem functions. In the present paper, we call for researchers especially from countries and ecoregions that remain underrepresented in the majority of soil biodiversity assessments to join us. Together we will be able to provide science-based evidence to support soil biodiversity conservation and functioning of terrestrial ecosystems.

ecology↗

bioRxiv2022

Globally invariant metabolism but density-diversity mismatch in springtails

Soil life supports the functioning and biodiversity of terrestrial ecosystems1, 2. Springtails (Collembola) are among the most abundant soil animals regulating soil fertility and flow of energy through above- and belowground food webs3-5. However, the global distribution of springtail diversity and density, and how these relate to energy fluxes remains unknown. Here, using a global dataset collected from 2,470 sites, we estimate total soil springtail biomass at 29 Mt carbon (threefold higher than wild terrestrial vertebrates6) and record peak densities up to 2 million individuals per m{superscript 2} in the Arctic. Despite a 20-fold biomass difference between tundra and the tropics, springtail energy use (community metabolism) remains similar across the latitudinal gradient, owing to the increase in temperature. Neither springtail density nor community metabolism were predicted by local species richness, which was highest in the tropics, but comparably high in some temperate forests and even tundra. Changes in springtail activity may emerge from latitudinal gradients in temperature, predation7, 8, and resource limitation7, 9, 10 in soil communities. Contrasting temperature responses of biomass, diversity and activity of springtail communities suggest that climate warming will alter fundamental soil biodiversity metrics in different directions, potentially restructuring terrestrial food webs and affecting major soil functions.

ecology↗

bioRxiv2021

Predictive regulation of cardiovascular system on emergence of auditory-motor interaction in young infants

Humans develop auditory-motor interaction to produce a variety of rhythmic sounds using body movements, which are often produced and amplified with tools, such as drumming. The extended production of sounds allows us to express a wide range of emotions, accompanied by physiological changes. According to previous studies, even young infants enhance movements in response to auditory feedback. However, their exhibition of physiological adaptation on emergence of auditory-motor interaction is unclear. We investigated the movement and cardiac changes associated with auditory feedback to spontaneous limb movements in 3-month-old infants. The results showed that infants increased the frequency of limb movements inducing auditory feedback, while they exhibited the more regular rhythm of the limb movements. Furthermore, heart rate increase associated with the limb movement was first inhibited immediately after the timing of the auditory feedback, which may reflect sustained attention to the auditory stimuli. Then, through auditory-motor experience, the heart rate increase was inhibited even prior to the auditory feedback, leading to suppression on the peak intensity of the heart rate increase. These findings suggest that young infants regulate the cardiovascular system as well as limb movements in anticipation of the auditory feedback. The anticipatory regulation associated with movement and attentional changes may contribute to reduced cardiovascular stress in auditory-motor interaction, and provide a developmental basis for more sophisticated goal-directed behavior of producing rhythmic sounds.

neuroscience↗

bioRxiv2021

HSD11β1 promotes EMT-mediated breast cancer metastasis

Abnormal biosyntheses of steroid hormones and dysregulation of steroid hormone receptors contribute to breast cancer metastasis but the mechanisms of that are poorly understand. Here we report a stress hormone producing enzyme, Hydroxysteroid (11-Beta) Dehydrogenase 1 (HSD11{beta}1) promotes breast cancer metastasis. HSD11{beta}1 was ectopically expressed in seventy-one percent of triple-negative breast tumors and correlated with shorter overall survival. HSD11{beta}1 significantly promoted breast cancer metastasis through induction of epithelial-to-mesenchymal transition (EMT); conversely, pharmacologic and genetic inhibition of HSD11{beta}1 suppressed metastatic progression of breast cancer cells. Moreover, 11-hydroxyprogesterone (11-OHP) whom HSD11{beta}1 produced in breast cancer cells, conferred metastatic properties on non-metastatic breast cancer cells through induction of EMT. We identified Peroxisome Proliferator-activated Receptor Alpha (PPAR-) as essential for both HSD11{beta}1 and 11OHP-driven EMT. Knockdown of PPAR- induced MET on HSD11{beta}1-expressing breast cancer cells. Taken together, HSD11{beta}1 promotes breast cancer metastasis and would be a novel target for suppressing breast cancer metastasis.

cancer biology↗

bioRxiv2021

Phosphatidylglycerol synthesis facilitates plastid gene expression and light induction of nuclear photosynthetic genes

Phosphatidylglycerol (PG) is the only major phospholipid in the thylakoid membrane of chloroplasts. PG is essential for photosynthesis and loss of PG in Arabidopsis thaliana results in severe defects of growth and chloroplast development with decreased chlorophyll accumulation, impaired thylakoid formation, and downregulation of photosynthesis-associated genes encoded in nuclear and plastid genomes. However, how the absence of PG affects the gene expression and plant growth remains unclear. To elucidate this mechanism, we investigated the growth and transcriptional profiles of a PG-deficient Arabidopsis mutant pgp1-2 under various light conditions. Microarray analysis demonstrated that reactive oxygen species-responsive genes were upregulated in pgp1-2. Decreased growth light did not alleviated the impaired leaf development and the downregulation of photosynthesis-associated genes in pgp1-2, indicating limited impacts of photooxidative stress on the defects of pgp1-2. Illumination to dark-adapted pgp1-2 triggered downregulation of photosynthesis-associated nuclear-encoded genes (PhANGs), while plastid-encoded genes were constantly suppressed. Overexpression of GOLDEN2-LIKE1 (GLK1), a transcription factor regulating chloroplast development, in pgp1-2 upregulated PhANGs but not plastid-encoded genes along with chlorophyll accumulation. Our data suggest a broad impact of PG biosynthesis on nuclear-encoded genes partially via GLK1 and a specific involvement of this lipid in the plastid gene expression and plant development.

plant biology↗

bioRxiv2021

LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec

Viral infection involves a large number of protein-protein interactions (PPIs) between human and virus. The PPIs range from the initial binding of viral coat proteins to host membrane receptors to the hijacking of host transcription machinery. However, few interspecies PPIs have been identified, because experimental methods including mass spectrometry are time-consuming and expensive, and molecular dynamic simulation is limited only to the proteins whose 3D structures are solved. Sequence-based machine learning methods are expected to overcome these problems. We have first developed the LSTM model with word2vec to predict PPIs between human and virus, named LSTM-PHV, by using amino acid sequences alone. The LSTM-PHV effectively learnt the training data with a highly imbalanced ratio of positive to negative samples and achieved an AUC of 0.976 with an accuracy of 98.4% using 5-fold cross-validation. By using independent test dataset, we compared the LSTM-PHV with existing state-of-the-art PPI predictors including DeepViral. In predicting PPIs between human and unknown or new virus, the LSTM-PHV presented higher performance than the existing predictors when they were trained by multiple host protein-including datasets. LSTM-PHV learnt multiple host protein sequence contexts more efficiently than the DeepViral. Interestingly, learning of only sequence contexts as words presented remarkably high performances. Use of uniform manifold approximation and projection demonstrated that the LSTM-PHV clearly distinguished the positive PPI samples from the negative ones. We presented the LSTM-PHV online web server that is freely available at http://kurata35.bio.kyutech.ac.jp/.

bioinformatics↗