An All-in-One Vehicle Type and License Plate Recognition System Using YOLOv4.

In smart surveillance and urban mobility applications, camera-equipped embedded platforms with deep learning technology have demonstrated applicability and effectiveness in identifying various targets. These use cases can be found in a variety of contexts and locations. It is critical to collect relevant data from the location where the application will be deployed. In this paper, we propose an integrated vehicle type and license plate recognition system using YOLOv4, which consists of vehicle type detection, license plate detection, and license plate character detection to better support the context of Korean vehicles in multilane highway and urban environments. Using our dataset of one to four multilane images, our system detected six vehicle classes and license plates with mAP of 98.0%, 94.0%, 97.1%, and 84.6%, respectively. On our dataset and a publicly available open dataset, our system demonstrated mAP of 99.3% and 99.4% for the detected license plates, respectively. From 4K high-resolution images, our system was able to detect minuscule license plates as small as 100 pixels wide. We believe that our system can be used in densely populated regions to address the high demands for enhanced visual sensitivity in smart cities and Internet-of-Things.

The transcriptional landscape and mutational profile of lung adenocarcinoma.

All cancers harbor molecular alterations in their genomes. The transcriptional consequences of these somatic mutations have not yet been comprehensively explored in lung cancer. Here we present the first large scale RNA sequencing study of lung adenocarcinoma, demonstrating its power to identify somatic point mutations as well as transcriptional variants such as gene fusions, alternative splicing events, and expression outliers. Our results reveal the genetic basis of 200 lung adenocarcinomas in Koreans including deep characterization of 87 surgical specimens by transcriptome sequencing. We identified driver somatic mutations in cancer genes including EGFR, KRAS, NRAS, BRAF, PIK3CA, MET, and CTNNB1. Candidates for novel driver mutations were also identified in genes newly implicated in lung adenocarcinoma such as LMTK2, ARID1A, NOTCH2, and SMARCA4. We found 45 fusion genes, eight of which were chimeric tyrosine kinases involving ALK, RET, ROS1, FGFR2, AXL, and PDGFRA. Among 17 recurrent alternative splicing events, we identified exon 14 skipping in the proto-oncogene MET as highly likely to be a cancer driver. The number of somatic mutations and expression outliers varied markedly between individual cancers and was strongly correlated with smoking history of patients. We identified genomic blocks within which gene expression levels were consistently increased or decreased that could be explained by copy number alterations in samples. We also found an association between lymph node metastasis and somatic mutations in TP53. These findings broaden our understanding of lung adenocarcinoma and may also lead to new diagnostic and therapeutic approaches.

FX: an RNA-Seq analysis tool on the cloud.

UNLABELLED: FX is an RNA-Seq analysis tool, which runs in parallel on cloud computing infrastructure, for the estimation of gene expression levels and genomic variant calling. In the mapping of short RNA-Seq reads, FX uses a transcriptome-based reference primarily, generated from ~160 000 mRNA sequences from RefSeq, UCSC and Ensembl databases. This approach reduces the misalignment of reads originating from splicing junctions. Unmapped reads not aligned on known transcripts are then mapped on the human genome reference. FX allows analysis of RNA-Seq data on cloud computing infrastructures, supporting access through a user-friendly web interface. AVAILABILITY: FX is freely available on the web at (http://fx.gmi.ac.kr), and can be installed on local Hadoop clusters. Guidance for the installation and operation of FX can be found under the 'Documentation' menu on the website. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

TIARA: a database for accurate analysis of multiple personal genomes based on cross-technology.

High-throughput genomic technologies have been used to explore personal human genomes for the past few years. Although the integration of technologies is important for high-accuracy detection of personal genomic variations, no databases have been prepared to systematically archive genomes and to facilitate the comparison of personal genomic data sets prepared using a variety of experimental platforms. We describe here the Total Integrated Archive of Short-Read and Array (TIARA; http://tiara.gmi.ac.kr) database, which contains personal genomic information obtained from next generation sequencing (NGS) techniques and ultra-high-resolution comparative genomic hybridization (CGH) arrays. This database improves the accuracy of detecting personal genomic variations, such as SNPs, short indels and structural variants (SVs). At present, 36 individual genomes have been archived and may be displayed in the database. TIARA supports a user-friendly genome browser, which retrieves read-depths (RDs) and log2 ratios from NGS and CGH arrays, respectively. In addition, this database provides information on all genomic variants and the raw data, including short reads and feature-level CGH data, through anonymous file transfer protocol. More personal genomes will be archived as more individuals are analyzed by NGS or CGH array. TIARA provides a new approach to the accurate interpretation of personal genomes for genome research.

xCyp26c Induced by Inhibition of BMP Signaling Is Involved in Anterior-Posterior Neural Patterning of Xenopus laevis.

Vertebrate neurogenesis requires inhibition of endogenous bone morphogenetic protein (BMP) signals in the ectoderm. Blocking of BMPs in animal cap explants causes the formation of anterior neural tissues as a default fate. To identify genes involved in the anterior neural specification, we analyzed gene expression profiles using a Xenopus Affymetrix Gene Chip after BMP-4 inhibition in animal cap explants. We found that the xCyp26c gene, encoding a retinoic acid (RA) degradation enzyme, was upregulated following inhibition of BMP signaling in early neuroectodermal cells. Whole-mount in situ hybridization analysis showed that xCyp26c expression started in the anterior region during the early neurula stage. Overexpression of xCyp26c weakly induced neural genes in animal cap explants. xCyp26c abolished the expression of all trans-/cis-RA-induced posterior genes, but not basic FGF-induced posterior genes. Depletion of xCyp26c by morpholino-oligonucleotides suppressed the normal formation of the axis and head, indicating that xCyp26c plays a critical role in the specification of anterior neural tissue in whole embryos. In animal cap explants, however, xCyp26c morpholinos did not alter anterior-to-posterior neural tissue formation. Together, these results suggest that xCyp26c plays a specific role in anterior-posterior (A-P) neural patterning of Xenopus embryos.

TIARA genome database: update 2013.

The Total Integrated Archive of short-Read and Array (TIARA; http://tiara.gmi.ac.kr) database stores and integrates human genome data generated from multiple technologies including next-generation sequencing and high-resolution comparative genomic hybridization array. The TIARA genome browser is a powerful tool for the analysis of personal genomic information by exploring genomic variants such as SNPs, indels and structural variants simultaneously. As of September 2012, the TIARA database provides raw data and variant information for 13 sequenced whole genomes, 16 sequenced transcriptomes and 33 high resolution array assays. Sequencing reads are available at a depth of ~30× for whole genomes and 50× for transcriptomes. Information on genomic variants includes a total of ~9.56 million SNPs, 23 025 of which are non-synonymous SNPs, and ~1.19 million indels. In this update, by adding high coverage sequencing of additional human individuals, the TIARA genome database now provides an extensive record of rare variants in humans. Following TIARA's fundamentally integrative approach, new transcriptome sequencing data are matched with whole-genome sequencing data in the genome browser. Users can here observe, for example, the expression levels of human genes with allele-specific quantification. Improvements to the TIARA genome browser include the intuitive display of new complex and large-scale data sets.

RNA-Seq analysis of frontal cortex and cerebellum from 5XFAD mice at early stage of disease pathology.

The pathogenesis of Alzheimer's disease (AD), especially the early events of AD pathology, remains unknown because of the complexity of AD and limitation of analysis methods. Transcriptome analysis has provided comprehensive insights to investigate the complex cellular activity in brain, but the transcriptome profiles from AD patients with microarray have generated discordant results. Here, for the first time, we performed transcriptome analysis of frontal cortex and cerebellum in 7-week-old 5XFAD transgenic mice (before extracellular amyloid plaque deposits) using high-throughput RNA-Seq analysis. Specific functional annotations were identified with differentially expressed genes (DEGs) of frontal cortex (a typically vulnerable region of AD pathology) and cerebellum (a typically non-vulnerable region of AD pathology). Cardiovascular disease-related genes were significantly found in down-regulated DEGs of frontal cortex, and mitochondrial dysfunction-related genes were evident in down-regulated DEGs of cerebellum. Additionally, we found RNA variants at the nucleotide level in transgenic mice compared with non-transgenic mice. Our results indicate that both frontal cortex and cerebellum in 5XFAD transgenic mice show specific pathological processes in the early pathophysiology of AD.

xCITED2 Induces Neural Genes in Animal Cap Explants of Xenopus Embryos.

Neural tissue is arisen from presumptive ectoderm via inhibition of bone morphogenetic protein (BMP) signaling during Xenopus early development. Previous studies demonstrate that ectopic expression of dominant negative BMP4 receptor (DNBR) produces neural tissue in animal cap explants (AC) and also increases the expression level of various genes involved in neurogenesis. To investigate detail mechanism of neurogenesis in transcriptional level, we analyzed RNAs increased by DNBR using total RNA sequencing analysis and identified several candidate genes. Among them, xCITED2 (Xenopus CBP/p300-interacting transcription activator) was induced 4.6 fold by DNBR and preferentially expressed in neural tissues at tadpole stage. Ectopic expression of xCITED2 induced anterior neural genes without mesoderm induction and reduced BMP downstream genes, an eye specific marker and posterior neural marker. Taken together, these results suggest that xCITED2 may have a role in the differentiation of anterior neural tissue during Xenopus early development.

Extensive genomic and transcriptional diversity identified through massively parallel DNA and RNA sequencing of eighteen Korean individuals.

Massively parallel sequencing technologies have identified a broad spectrum of human genome diversity. Here we deep sequenced and correlated 18 genomes and 17 transcriptomes of unrelated Korean individuals. This has allowed us to construct a genome-wide map of common and rare variants and also identify variants formed during DNA-RNA transcription. We identified 9.56 million genomic variants, 23.2% of which appear to be previously unidentified. From transcriptome sequencing, we discovered 4,414 transcripts not previously annotated. Finally, we revealed 1,809 sites of transcriptional base modification, where the transcriptional landscape is different from the corresponding genomic sequences, and 580 sites of allele-specific expression. Our findings suggest that a considerable number of unexplored genomic variants still remain to be identified in the human genome, and that the integrated analysis of genome and transcriptome sequencing is powerful for understanding the diversity and functional aspects of human genomic variants.

Swedish mutation within amyloid precursor protein modulates global gene expression towards the pathogenesis of Alzheimer's disease.

The Swedish mutation (K595N/M596L) of amyloid precursor protein (APP-swe) has been known to increase abnormal cleavage of cellular APP by Beta-secretase (BACE), which causes tau protein hyperphosphorylation and early-onset Alzheimer's disease (AD). Here, we analyzed the effect of APP-swe in global gene expression using deep transcriptome sequencing technique. We found 283 genes were down-regulated and 348 genes were up-regulated in APP-swe expressing H4-swe cells compared to H4 wild-type cells from a total of approximately 74 million reads of 38 base pairs from each transcriptome. Two independent mechanisms such as kinase and phosphatase signaling cascades leading hyperphosphorylation of tau protein were regulated by the expression of APP-swe. Expressions of catalytic subunit as well as several regulatory subunits of protein phosphatases 2A were decreased. In contrast, expressions of tau-phosphorylating glycogen synthase kinase 3ß (GSK-3ß), cyclin dependent kinase 5 (CDK5), and cAMP-dependent protein kinase A (PKA) catalytic subunit were increased. Moreover, the expression of AD-related Aquaporin 1 and presenilin 2 expression was regulated by APP-swe. Taken together, we propose that the expression of APP-swe modulates global gene expression directed to AD pathogenesis.