An application of convolutional neural networks with salient features for relation classification

Dashdorj, Zolzaya,Song, Min BioMed Central 2019 BMC bioinformatics Vol.20 No.10

<P><B>Background</B></P><P>Due to the advent of deep learning, the increasing number of studies in the biomedical domain has attracted much interest in feature extraction and classification tasks. In this research, we seek the best combination of feature set and hyperparameter setting of deep learning algorithms for relation classification. To this end, we incorporate an entity and relation extraction tool, PKDE4J to extract biomedical features (i.e., biomedical entities, relations) for the relation classification. We compared the chosen Convolutional Neural Networks (CNN) based classification model with the most widely used learning algorithms.</P><P><B>Results</B></P><P>Our CNN based classification model outperforms the most widely used supervised algorithms. We achieved a significant performance on binary classification with a weighted macro-average F1-score: 94.79% using pre-extracted relevant feature combinations. For multi-class classification, the weighted macro-average F1-score is estimated around 86.95%.</P><P><B>Conclusions</B></P><P>Our results suggest that our proposed CNN based model using the not only single feature as the raw text of the sentences of biomedical literature, but also coupling with multiple and highlighted features extracted from the biomedical sentences could improve the classification performance significantly. We offer hyperparameter tuning and optimization approaches for our proposed model to obtain optimal hyperparameters of the models with the best performance.</P>

Recent advances in biomedical applications of accelerator mass spectrometry

BioMed Central 2009 Journal of biomedical science Vol.16 No.1

<P>The use of radioisotopes has a long history in biomedical science, and the technique of accelerator mass spectrometry (AMS), an extremely sensitive nuclear physics technique for detection of very low-abundant, stable and long-lived isotopes, has now revolutionized high-sensitivity isotope detection in biomedical research, because it allows the direct determination of the amount of isotope in a sample rather than measuring its decay, and thus the quantitative analysis of the fate of the radiolabeled probes under the given conditions. Since AMS was first used in the early 90's for the analysis of biological samples containing enriched <SUP>14</SUP>C for toxicology and cancer research, the biomedical applications of AMS to date range from <I>in vitro </I>to <I>in vivo </I>studies, including the studies of 1) toxicant and drug metabolism, 2) neuroscience, 3) pharmacokinetics, and 4) nutrition and metabolism of endogenous molecules such as vitamins. In addition, a new drug development concept that relies on the ultrasensitivity of AMS, known as human microdosing, is being used to obtain early human metabolism information of candidate drugs. These various aspects of AMS are reviewed and a perspective on future applications of AMS to biomedical research is provided.</P>

SEQprocess: a modularized and customizable pipeline framework for NGS processing in R package

Joo, Taewoon,Choi, Ji-Hye,Lee, Ji-Hye,Park, So Eun,Jeon, Youngsic,Jung, Sae Hoon,Woo, Hyun Goo BioMed Central 2019 BMC bioinformatics Vol.20 No.-

<P><B>Backgrounds</B></P><P>Next-Generation Sequencing (NGS) is now widely used in biomedical research for various applications. Processing of NGS data requires multiple programs and customization of the processing pipelines according to the data platforms. However, rapid progress of the NGS applications and processing methods urgently require prompt update of the pipelines. Recent clinical applications of NGS technology such as cell-free DNA, cancer panel, or exosomal RNA sequencing data also require appropriate customization of the processing pipelines. Here, we developed SEQprocess, a highly extendable framework that can provide standard as well as customized pipelines for NGS data processing.</P><P><B>Results</B></P><P>SEQprocess was implemented in an R package with fully modularized steps for data processing that can be easily customized. Currently, six pre-customized pipelines are provided that can be easily executed by non-experts such as biomedical scientists, including the National Cancer Institute’s (NCI) Genomic Data Commons (GDC) pipelines as well as the popularly used pipelines for variant calling (e.g., GATK) and estimation of allele frequency, RNA abundance (e.g., TopHat2/Cufflink), or DNA copy numbers (e.g., Sequenza). In addition, optimized pipelines for the clinical sequencing from cell-free DNA or miR-Seq are also provided. The processed data were transformed into R package-compatible data type ‘ExpressionSet’ or ‘SummarizedExperiment’, which could facilitate subsequent data analysis within R environment. Finally, an automated report summarizing the processing steps are also provided to ensure reproducibility of the NGS data analysis.</P><P><B>Conclusion</B></P><P>SEQprocess provides a highly extendable and R compatible framework that can manage customized and reproducible pipelines for handling multiple legacy NGS processing tools.</P>

CoMAGC: a corpus with multi-faceted annotations of gene-cancer relations

Lee, Hee-Jin,Shim, Sang-Hyung,Song, Mi-Ryoung,Lee, Hyunju,Park, Jong C BioMed Central 2013 BMC bioinformatics Vol.14 No.-

<P><B>Background</B></P><P>In order to access the large amount of information in biomedical literature about genes implicated in various cancers both efficiently and accurately, the aid of text mining (TM) systems is invaluable. Current TM systems do target either gene-cancer relations or biological processes involving genes and cancers, but the former type produces information not comprehensive enough to explain how a gene affects a cancer, and the latter does not provide a concise summary of gene-cancer relations.</P><P><B>Results</B></P><P>In this paper, we present a corpus for the development of TM systems that are specifically targeting gene-cancer relations but are still able to capture complex information in biomedical sentences. We describe CoMAGC, a corpus with multi-faceted annotations of gene-cancer relations. In CoMAGC, a piece of annotation is composed of four semantically orthogonal concepts that together express 1) how a gene changes, 2) how a cancer changes and 3) the causality between the gene and the cancer. The multi-faceted annotations are shown to have high inter-annotator agreement. In addition, we show that the annotations in CoMAGC allow us to infer the prospective roles of genes in cancers and to classify the genes into three classes according to the inferred roles. We encode the mapping between multi-faceted annotations and gene classes into 10 inference rules. The inference rules produce results with high accuracy as measured against human annotations. CoMAGC consists of 821 sentences on prostate, breast and ovarian cancers. Currently, we deal with changes in gene expression levels among other types of gene changes. The corpus is available at http://biopathway.org/CoMAGCunder the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0).</P><P><B>Conclusions</B></P><P>The corpus will be an important resource for the development of advanced TM systems on gene-cancer relations.</P>

CollaboNet: collaboration of deep neural networks for biomedical named entity recognition

Yoon, Wonjin,So, Chan Ho,Lee, Jinhyuk,Kang, Jaewoo BioMed Central 2019 BMC bioinformatics Vol.20 No.10

<P><B>Background</B></P><P>Finding biomedical named entities is one of the most essential tasks in biomedical text mining. Recently, deep learning-based approaches have been applied to biomedical named entity recognition (BioNER) and showed promising results. However, as deep learning approaches need an abundant amount of training data, a lack of data can hinder performance. BioNER datasets are scarce resources and each dataset covers only a small subset of entity types. Furthermore, many bio entities are polysemous, which is one of the major obstacles in named entity recognition.</P><P><B>Results</B></P><P>To address the lack of data and the entity type misclassification problem, we propose CollaboNet which utilizes a combination of multiple NER models. In CollaboNet, models trained on a different dataset are connected to each other so that a target model obtains information from other collaborator models to reduce false positives. Every model is an expert on their target entity type and takes turns serving as a target and a collaborator model during training time. The experimental results show that CollaboNet can be used to greatly reduce the number of false positives and misclassified entities including polysemous words. CollaboNet achieved state-of-the-art performance in terms of precision, recall and F1 score.</P><P><B>Conclusions</B></P><P>We demonstrated the benefits of combining multiple models for BioNER. Our model has successfully reduced the number of misclassified entities and improved the performance by leveraging multiple datasets annotated for different entity types. Given the state-of-the-art performance of our model, we believe that CollaboNet can improve the accuracy of downstream biomedical text mining applications such as bio-entity relation extraction.</P>

MKEM: a Multi-level Knowledge Emergence Model for mining undiscovered public knowledge

Ijaz, Ali Z,Song, Min,Lee, Doheon BioMed Central 2010 BMC bioinformatics Vol.11 No.suppl2

<P><B>Background</B></P><P>Since Swanson proposed the Undiscovered Public Knowledge (UPK) model, there have been many approaches to uncover UPK by mining the biomedical literature. These earlier works, however, required substantial manual intervention to reduce the number of possible connections and are mainly applied to disease-effect relation. With the advancement in biomedical science, it has become imperative to extract and combine information from multiple disjoint researches, studies and articles to infer new hypotheses and expand knowledge.</P><P><B>Methods</B></P><P>We propose MKEM, a Multi-level Knowledge Emergence Model, to discover implicit relationships using Natural Language Processing techniques such as Link Grammar and Ontologies such as Unified Medical Language System (UMLS) MetaMap. The contribution of MKEM is as follows: First, we propose a flexible knowledge emergence model to extract implicit relationships across different levels such as molecular level for gene and protein and Phenomic level for disease and treatment. Second, we employ MetaMap for tagging biological concepts. Third, we provide an empirical and systematic approach to discover novel relationships.</P><P><B>Results</B></P><P>We applied our system on 5000 abstracts downloaded from PubMed database. We performed the performance evaluation as a gold standard is not yet available. Our system performed with a good precision and recall and we generated 24 hypotheses.</P><P><B>Conclusions</B></P><P>Our experiments show that MKEM is a powerful tool to discover hidden relationships residing in extracted entities that were represented by our Substance-Effect-Process-Disease-Body Part (SEPDB) model. </P>

Self-training in significance space of support vectors for imbalanced biomedical event data

Munkhdalai, Tsendsuren,Namsrai, Oyun-Erdene,Ryu, Keun Ho BioMed Central 2015 BMC bioinformatics Vol.16 No.suppl7

<P><B>Background</B></P><P>Pairwise relationships extracted from biomedical literature are insufficient in formulating biomolecular interactions. Extraction of complex relations (namely, biomedical events) has become the main focus of the text-mining community. However, there are two critical issues that are seldom dealt with by existing systems. First, an annotated corpus for training a prediction model is highly imbalanced. Second, supervised models trained on only a single annotated corpus can limit system performance. Fortunately, there is a large pool of unlabeled data containing much of the domain background that one can exploit.</P><P><B>Results</B></P><P>In this study, we develop a new semi-supervised learning method to address the issues outlined above. The proposed algorithm efficiently exploits the unlabeled data to leverage system performance. We furthermore extend our algorithm to a two-phase learning framework. The first phase balances the training data for initial model induction. The second phase incorporates domain knowledge into the event extraction model. The effectiveness of our method is evaluated on the Genia event extraction corpus and a PubMed document pool. Our method can identify a small subset of the majority class, which is sufficient for building a well-generalized prediction model. It outperforms the traditional self-training algorithm in terms of f-measure. Our model, based on the training data and the unlabeled data pool, achieves comparable performance to the state-of-the-art systems that are trained on a larger annotated set consisting of training and evaluation data.</P>

Acquisition of resistance to avian leukosis virus subgroup B through mutations on <i>tvb</i> cysteine-rich domains in DF-1 chicken fibroblasts

Lee, Hong Jo,Lee, Kyung Youn,Park, Young Hyun,Choi, Hee Jung,Yao, Yongxiu,Nair, Venugopal,Han, Jae Yong BioMed Central 2017 VETERINARY RESEARCH Vol.48 No.-

<P>Avian leukosis virus (ALV) is a retrovirus that causes tumors in avian species, and its vertical and horizontal transmission in poultry flocks results in enormous economic losses. Despite the discovery of specific host receptors, there have been few reports on the modulation of viral susceptibility via genetic modification. We therefore engineered acquired resistance to ALV subgroup B using CRISPR/Cas9-mediated genome editing technology in DF-1 chicken fibroblasts. Using this method, we efficiently modified the tumor virus locus B (<I>tvb</I>) gene, encoding the TVB receptor, which is essential for ALV subgroup B entry into host cells. By expanding individual DF-1 clones, we established that artificially generated premature stop codons in the cysteine-rich domain (CRD) of TVB receptor confer resistance to ALV subgroup B. Furthermore, we found that a cysteine residue (C80) of CRD2 plays a crucial role in ALV subgroup B entry. These results suggest that CRISPR/Cas9-mediated genome editing can be used to efficiently modify avian cells and establish novel chicken cell lines with resistance to viral infection.</P><P><B>Electronic supplementary material</B></P><P>The online version of this article (doi:10.1186/s13567-017-0454-1) contains supplementary material, which is available to authorized users.</P>

<i>Cordyceps bassiana</i> inhibits smooth muscle cell proliferation via the ERK1/2 MAPK signaling pathway

Jin, Enze,Han, Seongho,Son, Mina,Kim, Sung-Whan BioMed Central 2016 Cellular & molecular biology letters Vol.21 No.-

<P><I>Cordyceps</I> belongs to a genus of acormycete fungi and is known to exhibit various pharmacological effects. The aim of this study was to investigate the effect of <I>Cordyceps</I> species on the proliferation of vascular smooth muscle cells (VSMC) and their underlying molecular mechanism. A cell proliferation assay showed that <I>Cordyceps bassiana</I> ethanol extract (CBEE) significantly inhibited VSMC proliferation. In addition, neointimal formation was significantly reduced by treatment with CBEE in the carotid artery of balloon-injured rats. We also investigated the effects of CBEE on the extracellular signal-regulated kinase (ERK) signal pathway. Western blot analysis revealed increased ERK 1/2 phosphorylation in VSMCs treated with CBEE. Pretreatment with U0126 completely abrogated CBEE-induced ERK 1/2 phosphorylation. In conclusion, CBEE exhibited anti-proliferative properties that affected VSMCs through the ERK1/2 MAPK signaling pathway. Our data may elucidate the inhibitory mechanism of this natural product<I>.</I></P>

Peptidyl arginine deiminase type IV ( <i>PADI4</i> ) haplotypes interact with shared epitope regardless of anti-cyclic citrullinated peptide antibody or erosive joint status in rheumatoid arthritis: a case control study

Bang, So-Young,Han, Tae-Un,Choi, Chan-Bum,Sung, Yoon-Kyoung,Bae, Sang-Cheol,Kang, Changwon BioMed Central 2010 ARTHRITIS RESEARCH AND THERAPY Vol.12 No.3

<P><B>Introduction</B></P><P>Anti-cyclic citrullinated peptide autoantibodies (anti-CCP) are the most specific serologic marker for rheumatoid arthritis (RA). Genetic polymorphisms in a citrullinating (or deiminating) enzyme, peptidyl arginine deiminase type IV (PADI4) have been reproducibly associated with RA susceptibility in several populations. We investigated whether <I>PADI4 </I>polymorphisms contribute to anti-CCP-negative as well as -positive RA, whether they influence disease severity (erosive joint status), and whether they interact with two major risk factors for RA, Human Leukocyte Antigen-DRB1 <I>(HLA-DRB1</I>) shared epitope (SE) alleles and smoking, depending on anti-CCP and erosive joint status.</P><P><B>Methods</B></P><P>All 2,317 unrelated Korean subjects including 1,313 patients with RA and 1,004 unaffected controls were genotyped for three nonsynonymous (padi4_89, padi4_90, and padi4_92) and one synonymous (padi4_104) single-nucleotide polymorphisms (SNPs) in <I>PADI4 </I>and for <I>HLA-DRB1 </I>by direct DNA sequence analysis. Odds ratios (OR) were calculated by multivariate logistic regression. Interaction was evaluated by attributable proportions (AP), with 95% confidence intervals (CI).</P><P><B>Results</B></P><P>A functional haplotype of the three fully correlated nonsynonymous SNPs in <I>PADI4 </I>was significantly associated with susceptibility to not only anti-CCP-positive (adjusted OR 1.73, 95% CI 1.34 to 2.23) but also -negative RA (adjusted OR 1.75, 95% CI 1.15 to 2.68). A strong association with both non-erosive (adjusted OR 1.62, 95% CI 1.29 to 2.05) and erosive RA (adjusted OR 1.62, 95% CI 1.14 to 2.31) was observed for <I>PADI4 </I>haplotype. Gene-gene interactions between the homozygous RA-risk <I>PADI4 </I>haplotype and SE alleles were significant in both anti-CCP-positive (AP 0.45, 95% CI 0.20 to 0.71) and -negative RA (AP 0.61, 95% CI 0.29 to 0.92). Theses interactions were also observed for both non-erosive (AP 0.48, 95% CI 0.25 to 0.72) and erosive RA (AP 0.46, 95% CI 0.14 to 0.78). In contrast, no interaction was observed between smoking and <I>PADI4 </I>polymorphisms.</P><P><B>Conclusions</B></P><P>A haplotype of nonsynonymous SNPs in <I>PADI4 </I>contributes to development of RA regardless of anti-CCP or erosive joint status. The homozygous <I>PADI4 </I>haplotype contribution is affected by gene-gene interactions with <I>HLA-DRB1 </I>SE alleles.</P>