Biology subjects

Miyano, S.

Publications and source records attributed to Miyano, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Chromatin landscape and epigenetic heterogeneity of acute myeloid leukemia

Acute myeloid leukemia (AML) is an aggressive hematologic cancer characterized by proliferation of immature myeloblasts. It shows profound molecular heterogeneity, which has been primarily studied through genetic abnormalities, providing the basis for disease classification, prognostication, and therapeutic choice. However, genetic factors alone may not fully explain AML pathogenesis and diversity, while leaving the role of abnormal epigenome, particularly chromatin state, largely unexplored in a large cohort of patients. Here we show that AML is classified into 16 subgroups with distinct chromatin accessibility profiles based on ATAC-seq in 1,563 AML cases, including novel AML subgroups not previously recognized in conventional genomic classifications. By integrating multi-omics analyses of genome, transcriptome, and major histone marks, we show that these epigenetic subgroups exhibit unique features in clinical presentation, gene mutations, differentiation states, gene expression, and super-enhancer profiles, which are validated across independent cohorts. Single-cell sequencing demonstrates the presence of subgroup-specific ATAC signatures that are shared by all leukemic cells, confirming the definitive role of the epigenome in the ATAC-based classification. Mechanistically, each subgroup is associated with a distinct gene regulatory network centered on key transcription factors, where subgroup-specific super-enhancers play a pivotal role. These ATAC subgroups also have prognostic significance independent of genomic classification, and help reveal unexpected drug sensitivities. In summary, ATAC-based chromatin profiling in this large sample set, combined with multi-omics data, provides new insights into AML pathogenesis beyond genomic profiling and also serves as an invaluable resource for AML research.

cancer biology↗

bioRxiv2025

Cohesin constrains histone modification-driven chromatin dynamics

Gene expressions are regulated by an interplay between epigenetics and spatial genome organization, the deregulation of which has been implicated in the development of cancers, including myeloid neoplasms. However, it is unclear how they coordinately contribute to normal and malignant hematopoiesis. Here, we show that simultaneous dysregulations of histone modifications and chromatin structures caused by mutations of the epigenetic modulator Asxl1 and cohesin subunit Stag2 cooperatively induce ectopic interactions between polycomb-regulated promoters and active enhancers, leading to the aberrant upregulation of hematopoietic stem cell related genes and development of myelodysplastic syndromes (MDS). De-repression of polycomb-regulated genes induces their translocation to the transcriptionally active loci, where active promoters and enhancers are assembled, in the absence of Stag2-mediated chromatin organization. Our findings revealed that cohesin counteracts histone modification-driven chromatin conformations, manifesting the coordinate roles of histone modifiers and cohesin in regulating genome architectures and gene expressions to prevent malignant transformation.

cancer biology↗

bioRxiv2021

Identification of Bacteriophages Using Deep Representation Model with Pre-training

MotivationBacteriophages/Phages are the viruses that infect and replicate within bacteria and archaea, and rich in human body. To investigate the relationship between phages and microbial communities, the identification of phages from metagenome sequences is the first step. Currently, there are two main methods for identifying phages: database-based (alignment-based) methods and alignment-free methods. Database-based methods typically use a large number of sequences as references; alignment-free methods usually learn the features of the sequences with machine learning and deep learning models. ResultsWe propose INHERIT which uses a deep representation learning model to integrate both database-based and alignment-free methods, combining the strengths of both. Pre-training is used as an alternative way of acquiring knowledge representations from existing databases, while the BERT-style deep learning framework retains the advantage of alignment-free methods. We compare INHERIT with four existing methods on a third-party benchmark dataset. Our experiments show that INHERIT achieves a better performance with the F1-score of 0.9932. In addition, we find that pre-training two species separately helps the non-alignment deep learning model make more accurate predictions. AvailabilityThe codes of INHERIT are now available in: https://github.com/Celestial-Bai/INHERIT. Contactyaozhong@ims.u-tokyo.ac.jp and imoto@hgc.jp Supplementary informationSupplementary data are available at BioRxiv online.

bioinformatics↗

bioRxiv2021

Combined landscape of single-nucleotide variants and copy-number alterations in clonal hematopoiesis

Implicated in the development of hematological malignancies (HM) and cardiovascular mortality, clonal hematopoiesis (CH) in apparently healthy individuals has been investigated by detecting either single-nucleotide variants and indels (SNVs/indels) or copy number alterations (CNAs), but not both. Here by combining targeted sequencing of 23 CH-related genes and array-based CNA detection of blood-derived DNA, we have delineated the landscape of CH-related SNVs/indels and CNAs in a general population of 11,234 individuals, including 672 with subsequent HM development. Both CH-related lesions significantly co-occurred, which combined, affected blood count, hypertension, and the mortality from HM and cardiovascular diseases depending on the total number of both lesions, highlighting the importance of detecting both lesions in the evaluation of CH.

cancer biology↗

bioRxiv2021

On the application of BERT models for nanopore methylation detection

MotivationDNA methylation is a common epigenetic modification, which is widely associated with various biological processes, such as gene expression, aging, and disease. Nanopore sequencing provides a promising methylation detection approach through monitoring abnormal signal shifts for detecting modified bases in target motif regions. Recently, model-based approaches, especially those with deep learning models, have achieved significant performance improvements on nanopore methylation detection. In this work, we explore using bidirectional encoder representations from transformers (BERT) for doing the task, which can provide non-recurrent neural structures for fast parallel computation. ResultsWe find original BERT architecture does not work as well as the bidirectional recurrent neural network (biRNN) on the nanopore methylation prediction task. Through further analysis, we observe recurrent patterns of positional-signal-shift in the context window surrounding target 5-methylcytosine (5mC) and N6-methyladenine (6mA) motifs. We propose a refined BERT with relative position representation and center hidden units concatenation, which takes account of task-specific characters into modeling. We perform systematic evaluations in-sample and cross-sample. The experiment results show that the refined BERT model can achieve competitive or even better results than the state-of-the-art biRNN model, while the model inference speed is about 6x faster. Besides, on the cross-sample evaluation of datasets from the different research groups, BERT models demonstrate a good generalization performance. AvailabilityThe source code and data are available at https://github.com/yaozhong/methBERT Contactyaozhong@ims.u-tokyo.ac.jp

bioinformatics↗

bioRxiv2020

RoDiCE: Robust differential protein co-expression analysis for cancer complexome

MotivationThe full spectrum of abnormalities in cancer-associated protein complexes remains largely unknown. Comparing the co-expression structure of each protein complex between tumor and healthy cells may provide insights regarding cancer-specific protein dysfunction. However, the technical limitations of mass spectrometry-based proteomics, including contamination with biological protein variants, causes noise that leads to non-negligible over- (or under-) estimating co-expression. ResultsWe propose a robust algorithm for identifying protein complex aberrations in cancer based on differential protein co-expression testing. Our method based on a copula is sufficient for improving identification accuracy with noisy data compared to conventional linear correlation-based approaches. As an application, we use large-scale proteomic data from renal cancer to show that important protein complexes, regulatory signaling pathways, and drug targets can be identified. The proposed approach surpasses traditional linear correlations to provide insights into higher order differential co-expression structures. Availability and Implementationhttps://github.com/ymatts/RoDiCE.

bioinformatics↗

bioRxiv2019

A multifunctional R package for identification of tumor-specific neoantigens

It is known that some mutated peptides, such as those resulting from missense mutations and frameshift insertions, can bind to the major histocompatibility complex and be presented to antitumor T-cells on the surface of a tumor cell. These peptides are termed neoantigen and it is important to understand this process for cancer immunotherapy. Here, we introduce an R package that can predict a list of potential neoantigens from a variety of mutations, which include not only somatic point mutations but insertions, deletions, and structural variants. Beyond the existing applications, this package is capable of attaching and reflecting several additional information, e.g., wild-type binding capability, allele specific RNA expression levels, single nucleotide polymorphism information, and combinations of mutations to filter out infeasible peptides as neoantigen. AvailabilityThe R package is available at http://github/hase62/Neoantimon.

bioinformatics↗

bioRxiv2019

A unified simulation model for understanding the diversity of cancer evolution

Because cancer evolution underlies the therapeutic difficulties of cancer, it is clinically important to understand the evolutionary dynamics of cancer. Thus far, a number of evolutionary processes have been proposed to be working in cancer evolution. However, there exists no simulation model that can describe the different evolutionary processes in a unified manner. In this study, we constructed a unified simulation model for describing the different evolutionary processes and performed sensitivity analysis on the model to determine the conditions in which cancer growth is driven by each of the different evolutionary processes. Our sensitivity analysis has successfully provided a series of novel insights into the evolutionary dynamics of cancer. For example, we found that, while a high neutral mutation rate shapes neutral intratumor heterogeneity (ITH) characterized by a fractal-like pattern, a stem cell hierarchy can also contribute to shaping neutral ITH by apparently increasing the mutation rate. Although It has been reported that the evolutionary principle shaping ITH shifts from selection to accumulation of neutral mutations during colorectal tumorigenesis, our simulation revealed the possibility that this evolutionary shift is triggered by drastic evolutionary events that occur in a a short time and confer a marked fitness increase on one or a few cells. This result helps us understand that each process works not separately but simultaneously and continuously as a series of phases of cancer evolution. Collectively, this study serves as a basis to understand in greater depth the diversity of cancer evolution.

cancer biology↗

bioRxiv2019

Nanopore base-calling from a perspective of instance segmentation

BackgroundNanopore sequencing is a rapidly developing third-generation sequencing technology, which can generate long nucleotide reads of molecules within a portable device in real time. Through detecting the change of ion currency signals during a DNA/RNA fragments pass through a nanopore, genotypes are determined. Currently, the accuracy of nanopore base-calling has a higher error rate than short-read base-calling. Through utilizing deep neural networks, the-state-of-the art nanopore base-callers achieve base-calling accuracy in a range from 85% to 95%.\n\nResultIn this work, we proposed a novel base-calling approach from a perspective of instance segmentation. Different from the previous sequence labeling approaches, we formulated the base-calling problem as a multi-label segmentation task. Meanwhile, we proposed a refined U-net model which we call UR-net that can model sequential dependencies for a one-dimensional segmentation task. The experiment results show that the proposed base-caller URnano achieves competitive results compared to recently proposed CTC-featured base-caller Chiron, on the same amount of training and test data for in-domain evaluation. Our results show that formulating the base-calling problem as a one-dimensional segmentation task is a promising approach.\n\nAvailabilityThe source code and data are available at https://github.com/yaozhong/URnano\n\nContactyaozhong@ims.u-tokyo.ac.jp\n\nSupplementary informationSupplementary data are available at attachment online.

bioinformatics↗

bioRxiv2019

Intra-bin structural variant detection for whole genome sequencing data using U-net

MotivationFor short-read sequencing, read-depth based structural variant (SV) callers are difficult to find single-nucleotide-resolution breakpoints due to the bin-size limitation. ResultsIn this paper, we present RDBKE to enhance the breakpoint resolution of read-depth SV callers using deep segmentation model UNet. We show that UNet can be trained with a small amount of data and applied for breakpoint enhancement both in-sample and cross-sample. On both simulation and real data, RDBKE significantly increases the number of SVs with more precise breakpoints. Availabilitysource code of RDBKE is available at https://github.com/yaozhong/deepIntraSV Contactyaozhong@ims.u-tokyo.ac.jp

bioinformatics↗

bioRxiv2019

Sensitivity analysis of agent-based simulation utilizing massively parallel computation and interactive data visualization

An essential step in the analysis of agent-based simulation is sensitivity analysis, which namely examines the dependency of parameter values on simulation results. Although a number of approaches have been proposed for sensitivity analysis, they still have limitations in exhaustivity and interpretability. In this study, we propose a novel methodology for sensitivity analysis of agent-based simulation, MASSIVE (Massively parallel Agent-based Simulations and Subsequent Interactive Visualization-based Exploration). MASSIVE takes a unique paradigm, which is completely different from those of sensitivity analysis methods developed so far, By combining massively parallel computation and interactive data visualization, MASSIVE enables us to inspect a broad parameter space intuitively. We demonstrated the utility of MASSIVE by its application to cancer evolution simulation, which successfully identified conditions that generate heterogeneous tumors. We believe that our approach would be a de facto standard for sensitivity analysis of agent-based simulation in an era of ever-growing computational technology. All the result form our MASSIVE analysis is available at https://www.hgc.jp/~niiyan/massive.

bioinformatics↗

bioRxiv2018

Comprehensive Analysis of Indels in Whole-genome Microsatellite Regions and Microsatellite Instability across 21 Cancer Types

Microsatellites are repeats of 1-6bp units and [~]10 million microsatellites have been identified across the human genome. Microsatellites are vulnerable to DNA mismatch errors, and have thus been used to detect cancers with mismatch repair deficiency. To reveal the mutational landscape of the microsatellite repeat regions at the genome level, we analyzed approximately 20.1 billion microsatellites in 2,717 whole genomes of pan-cancer samples across 21 tissue types. Firstly, we developed a new insertion and deletion caller (MIMcall) that takes into consideration the error patterns of different types of microsatellites. Among the 2,717 pan-cancer samples, our analysis identified 31 samples, including colorectal, uterus, and stomach cancers, with higher microsatellite mutation rate ([≥] 0.03), which we defined as microsatellite instability (MSI) cancers in genome-wide level. Next, we found 20 highly-mutated microsatellites that can be used to detect MSI cancers with high sensitivity. Third, we found that replication timing and DNA shape were significantly associated with mutation rates of the microsatellites. Analysis of germline variation of the microsatellites suggested that the amount of germline variations and somatic mutation rates were correlated. Lastly, analysis of mutations in mismatch repair genes showed that somatic SNVs and short indels had larger functional impact than germline mutations and structural variations. Our analysis provides a comprehensive picture of mutations in the microsatellite regions, and reveals possible causes of mutations, as well as provides a useful marker set for MSI detection.

cancer biology↗

bioRxiv2018

GIMLET: Identifying Biological Modulators in Context-Specific Gene Regulation Using Local Energy Statistics

The regulation of transcription factor activity dynamically changes across cellular conditions and disease subtypes. The identification of biological modulators contributing to context-specific gene regulation is one of the challenging tasks in systems biology, which is necessary to understand and control cellular responses across different genetic backgrounds and environmental conditions. Previous approaches for identifying biological modulators from gene expression data were restricted to the capturing of a particular type of a three-way dependency among a regulator, its target gene, and a modulator; these methods cannot describe the complex regulation structure, such as when multiple regulators, their target genes, and modulators are functionally related. Here, we propose a statistical method for identifying biological modulators by capturing multivariate local dependencies, based on energy statistics, which is a class of statistics based on distances. Subsequently, our method assigns a measure of statistical significance to each candidate modulator through a permutation test. We compared our approach with that of a leading competitor for identifying modulators, and illustrated its performance through both simulations and real data analysis. Our method, entitled genome-wide identification of modulators using local energy statistical test (GIMLET), is implemented with R ([≥] 3.2.2) and is available from github (https://github.com/tshimam/GIMLET).

bioinformatics↗

bioRxiv2018

ALPHLARD: a Bayesian method for analyzing HLA genes from whole genome sequence data

Although human leukocyte antigen (HLA) genotyping based on amplicon, whole exome sequence (WES), and RNA sequence data has been achieved in recent years, accurate genotyping from whole genome sequence (WGS) data remains a challenge due to the low depth. Furthermore, there is no method to identify the sequences of unknown HLA types not registered in HLA databases. We developed a Bayesian model, called ALPHLARD, that collects reads potentially generated from HLA genes and accurately determines a pair of HLA types for each of HLA-A, -B, -C, -DPA1, -DPB1, -DQA1, -DQB1, and -DRB1 genes at 6-digit resolution. Furthermore, ALPHLARD can detect rare germline variants not stored in HLA databases and call somatic mutations from paired normal and tumor sequence data. We illustrate the capability of ALPHLARD using 253 WES data and 25 WGS data from Illumina platforms. By comparing the results of HLA genotyping from SBT and amplicon sequencing methods, ALPHLARD achieved 98.8% for WES data and 98.5% for WGS data at 4-digit resolution. We also detected three somatic point mutations and one case of loss of heterozygosity in the HLA genes from the WGS data. ALPHLARD showed good performance for HLA genotyping even from low-coverage data. It also has a potential to detect rare germline variants and somatic mutations in HLA genes. It would help to fill in the current gaps in HLA reference databases and unveil the immunological significance of somatic mutations identified in HLA genes.

bioinformatics↗

bioRxiv2018

Immuno-genomic PanCancer Landscape Reveals Diverse Immune Escape Mechanisms and Immuno-Editing Histories

Immune reactions in the tumor micro-environment are one of the cancer hallmarks and emerging immune therapies have been proven effective in many types of cancer. To investigate cancer genome-immune interactions and the role of immuno-editing or immune escape mechanisms in cancer development, we analyzed 2,834 whole genomes and RNA-seq datasets across 31 distinct tumor types from the PanCancer Analysis of Whole Genomes (PCAWG) project with respect to key immuno-genomic aspects. We show that selective copy number changes in immune-related genes could contribute to immune escape. Furthermore, we developed an index of the immuno-editing history of each tumor sample based on the information of mutations in exonic regions and pseudogenes. Our immuno-genomic analyses of pan-cancer analyses have the potential to identify a subset of tumors with immunogenicity and diverse background or intrinsic pathways associated with their immune status and immuno-editing history.

genomics↗

bioRxiv2017

A framework for generating interactive reports for cancer genome analysis

SummaryWe introduce paplot, the software for generating dynamic reports that are frequently necessary in the post analytical phases of cancer genome studies. The \"interactive\" nature of the paplot-generated reports enables users to extract much richer information than that obtained from static graphs via most conventional visualization tools.\n\nAvailability and implementationThe python implementation for paplot (MIT license) is available at https://github.com/Genomon-Project/paplot. The documentation is at http://paplot-doc.readthedocs.io/en/latest/.\n\nContactyshira@hgc.jp

bioinformatics↗

bioRxiv2017

A comprehensive characterization of cis-acting splicing-associated variants in human cancer

Although many driver mutations are thought to promote carcinogenesis via abnormal splicing, the landscape of these splicing-associated variants (SAVs) remains unknown due to the complexity of splicing abnormalities. Here we developed a statistical framework to identify SAVs disrupting or newly creating splice site motifs and applied it to sequencing data from 8,976 samples across 31 cancer types. We constructed a catalog of 14,438 SAVs, approximately 50% of which consist of SAVs disrupting non-canonical splice sites (including the 3rd and 5th intronic bases of donor sites) or newly creating splice sites. Smoking-related signature substantially contributes to SAV generation. As many as 14.7% of samples harbor at least one SAVs in cancer-related genes, particularly in tumor suppressors. Importantly, in addition to previously reported intron retention, exon skipping or alternative splice site usage more frequently affected these genes. Our findings delineate a comprehensive portrait of SAVs, providing a basis for cancer precision medicine.

genomics↗

bioRxiv2017

Tumor subclonal progression model for cancer hallmark acquisition

Recent advances in the methods for reconstruction of cancer evolutionary trajectories opened up the prospects of deciphering the subclonal populations and their evolutionary architectures within cancer ecosystems. An important challenge of the cancer evolution studies is how to connect genetic aberrations in subclones to a clinically interpretable and actionable target in the subclones for individual patients. In this study, our aim is to develop a novel method for constructing a model of tumor subclonal progression in terms of cancer hallmark acquisition using multiregional sequencing data. We prepare a subclonal evolutionary tree inferred from variant allele frequencies and estimate pathway alteration probabilities from large-scale cohort genomic data. We then construct an evolutionary tree of pathway alterations that takes into account selectivity of pathway alterations via selectivity score. We show the effectiveness of our method on a dataset of clear cell renal cell carcinomas.

bioinformatics↗