Search

SUMMARYGenomic abnormalities, including structural variation (SV), copy number variation (CNV), single-nucleotide polymorphism (SNP), homogenously staining regions (HSR) and extrachromosomal DNA (ecDNA), are strongly associated with cancer, rare diseases and infertility. A robust technology to simultaneously detect these genomic abnormalities is highly desired for clinical diagnosis and basic research. In this study, we developed a simple and cost-effective method – multiple genetic abnormality sequencing (MGA-Seq) – to simultaneously detect SNPs, CNVs, SVs, ecDNA and HSRs in a single tube. This method has been successfully applied in both cancer cell lines and clinical tumour samples and revealed that focal amplification in tumour tissue is substantially heterogeneous. Notably, we delineated the architecture of focal amplification and the ecDNA network by MGA-Seq, which facilitated the exploration of the regulation of gene expression in ecDNA. This method could be extensively applied for diagnosis and may greatly facilitate the investigation of the genomic mechanism for genetic diseases.

The impressive functions of the brain rely on an extensive connectivity matrix between specific neurons, the architecture of which is frequently characterized by one brain nucleus/region connecting to multiple targets, either via collaterals of the same projection neuron or several, differentially specified neurons. Delineating the fine architecture of projection neuron subsets in a specific brain region could greatly facilitate its circuit, computational, and functional resolution. Here, we developed multiple fluorescent rabies viruses (RV) to delineate the fine organization of corticothalamic projection neuron subsets in the primary visual cortex (V1). By simultaneously retrograde labeling multiple distinct subsets of corticothalamic projection neurons in V1 from their target nuclei in thalamus (dLGN, LP, LD), we observed that V1-dLGN corticothalamic projection neurons were densely concentrated in layer VI, except for several sparsely scattered neurons in layer V, while V1-LP and V1-LD corticothalamic projection neurons were localized to both layers V and VI. Meanwhile, we observed a fraction of V1 corticothalamic projection neurons targeting two thalamic nuclei, which was further confirmed by fMOST whole-brain imaging. The multiple fluorescent RV tracing tools can be extensively applied to resolve the architecture of projection neuron subsets in certain brain regions, with a strong potential to delineate the computational and functional organization of these brain regions.

Knowledge graph embedding is studied to embed entities and relations of a knowledge graph into continuous vector spaces, which benefits a variety of real-world applications. Among existing solutions, translation-based models, which employ geometric translation to design score function, have drawn much attention. However, these models primarily concentrate on evidence from observing whether the triplets are plausible, and ignore the fact that the relation also implies certain semantic constraints on its subject or object entity. In this paper, we present a general framework for enhancing knowledge graph embedding with relational constraints (KRC). Specifically, we elaborately design the score function by encoding regularities between a relation and its arguments into the translation-based embedding space. Additionally, we propose a soft margin-based ranking loss for effectively training the KRC model, which characterizes different semantic distances between negative and positive triplets. Furthermore, we combine regularities with distributional representations to predict the missing triplets, which possesses certain robust guarantee. We evaluate our method on the task of knowledge graph completion. Extensive experiments show that KRC achieves substantial improvements against baselines.

Studies on the 3D genomics could facilitate the understanding of relationship between genome structure and functions. Based on different research purposes, a variety of chromatin three-dimensional conformation capture methods have been developed, greatly promoting the progress of three-dimensional genomics research. This review aims to systematically introduce and summarize various 3D genomics research methods. The methods are generally divided into three categories. The first category is based on proximity ligation and sequencing, including 3C, Hi-C, in situ Hi-C, capture Hi-C, single-cell Hi-C, DLO Hi-C, and ChIA-PET. The other category is not based on proximity ligation, such as ChIA-Drop, SPRITE, and GAM. The last category is based on microscopy, including seqFISH, CRISPRainbow, and SABER amplifies FISH. Finally, the perspective of conformation capture technologies is discussed.

The highly organized three-dimensional genome is crucial for gene transcription. However, it remains elusive how the order of the genome architecture related to its function. Here, we developed a single-cell Hi-C method and proposed TAD “degree of disorder” as a measure of genome organizational patterns, which is correlated with the chromatin epigenetic states, gene expression and co-regulation, and chromatin structure variability in individual cells. Upon Mycobacterium tuberculosis infection, NF-κB enters into the nucleus, binds to the target genome regions and initiates systematic chromatin conformation reorganization. Furthermore, we identified a remote NF-κB enriched enhancer promotes the expression of PD-L1 through chromatin loop, which could be a potential anti-tuberculosis and even anti-tumor therapeutic target. The integrated Hi-C, eQTL, and GWAS analysis depicted the atlas of the long-range target genes of tuberculosis susceptible loci. Among which SNP rs1873613 is located in the anchor of a dynamic chromatin loop with LRRK2, whose inhibitor AdoCbl could be an anti-tuberculosis drug candidate. Our study provides comprehensive resources for the 4D genome of immunocytes and sheds insights into the genome organization order and the coordinated gene transcription.

Distance metric learning aims to learn a metric with the similarity of samples. However, the increasing scalability and complexity of dataset or complex application brings about inevitable label noise, which frustrates the distance metric learning. In this paper, we propose a resampling scheme robust to label noise, Re-KISSME, based on Keep It Simple and Straightforward Metric (KISSME) learning method. Specifically, we consider the data structure and the priors of labels as two resampling factors to correct the observed distribution. By introducing the true similarity as latent variable, these two factors are integrated into a maximum likelihood estimation model. As a result, Re-KISSME can reason the underlying similarity of each pair and reduce the influence of label noise to estimate the metric matrix. Our model is solved by iterative algorithm with low computational cost. With synthetic label noise, the experiments on UCI datasets and two application datasets of person re-identification confirm the effectiveness of our proposal.

Automatic question answering (QA) system is the inevitable trend of future search engines. As the essential steps of QA, question classification and text retrieval both require algorithms to capture the semantic information and syntactic structure of natural language. This paper proposes dependency-based convolutional networks to learn a representation of sentences. First, we use dependency layer to map discrete word depth on the dependency tree of a sentence into continuous real space. Then, the mapping result serves as weight of word vectors and convolutional kernels are employed as feature extractors for further specific tasks. The method proposed allows convolutional networks to take the advantage of higher representational ability of dependency structure. Experiments involving three tasks including text classification, duplicate classification and text pairs ranking confirm the advantages of our model.

The impressive functions of the brain rely on an extensive connectivity matrix between specific neurons, the architecture of which is frequently characterized by one brain nucleus/region connecting to multiple targets, either via collaterals of the same projection neuron or several, differentially specified neurons. Delineating the fine architecture of projection neuron subsets in a specific brain region could greatly facilitate its circuit, computational, and functional resolution. Here, we developed multiple fluorescent rabies viruses (RV) to delineate the fine organization of corticothalamic projection neuron subsets in the primary visual cortex (V1). By simultaneously retrograde labeling multiple distinct subsets of corticothalamic projection neurons in V1 from their target nuclei in thalamus (dLGN, LP, LD), we observed that V1-dLGN corticothalamic projection neurons were densely concentrated in layer VI, except for several sparsely scattered neurons in layer V, while V1-LP and V1-LD corticothalamic projection neurons were localized to both layers V and VI. Meanwhile, we observed a fraction of V1 corticothalamic projection neurons targeting two thalamic nuclei, which was further confirmed by fMOST whole-brain imaging. The multiple fluorescent RV tracing tools can be extensively applied to resolve the architecture of projection neuron subsets in certain brain regions, with a strong potential to delineate the computational and functional organization of these brain regions.

Recognizing standard medical concepts in the colloquial text is significant for kinds of applications such as the medical question answering system. Recently, word-level neural network methods, which can learn complex informal expression features, achieved remarkable performance on this task. However, they have two main limitations: (1) Existing word-level methods cannot learn character structure features inside words and suffer from “Out-of-vocabulary” (OOV) words, which are common in noisy colloquial text. (2) Since these methods handle the normalization task as a classification issue, concept phrases are represented by category labels. Hence the word morphological information inside the concept is lost. In this work, we present a multi-task character-level attentional network model for medical concept normalization. Specifically, the character-level encoding scheme of our model can alleviate the OOV word problem. The attention mechanism can effectively exploit the word morphological information through multi-task training. It generates higher attention weights on domain-related positions in the text sequence, helping the downstream convolution focus on the characters that are related to medical concepts. To test our model, we first introduce a labeled Chinese dataset (overall 314,991 records) for this task. Other two real-world English datasets are also used. Our model outperforms state-of-the-art methods on all three datasets. Besides, by adding four types noises to the datasets, we validate the robustness of our model against common noises in the colloquial text.

The Ca2+ sensor proteins STIM1 and STIM2 are crucial elements of store-operated calcium entry (SOCE) in breast cancer cells. Increased SOCE activity may contribute to epithelial-mesenchymal transitions (EMT) and increase cell migration and invasion. However, the roles of STIM1 and STIM2 in TGF-β-induced EMT are still unclear. In this study, we demonstrate roles of STIMs in TGF-β-induced EMT in breast cancer cells. In particular, STIM1 and STIM2 expression affected TGF-β-induced EMT by mediating SOCE in MDA-MB-231 and MCF-7 breast cancer cells. The specific SOCE inhibitor YM58483 blocked TGF-β-induced EMT, and differing effects of STIM1 and STIM2 on TGF-β-induced EMT correlated with differing roles in SOCE. Finally, we showed that STIM2 is associated with non-store-operated calcium entry (non-SOCE) during TGF-β-induced EMT, whereas STIM1 is not. What's more, non-SOCE have a large possibility to be ROCE. In conclusion, STIM1 and STIM2 proteins play important roles in TGF-β-induced EMT and these effects are related to both SOCE and non-SOCE.

Background Stromal interaction molecule (STIM) 2 is a key calcium-sensing molecule that regulates the stabilization of calcium ions (Ca2+) and therefore regulates downstream Ca2+-associated signaling and cellular events. We hypothesized that STIM2 regulates epithelial-mesenchymal transition (EMT) to promote breast cancer metastasis. Methods We determined the effects of gain, loss, and rescue of STIM2 on cellular motility, levels of EMT-related proteins, and secretion of transforming growth factor-β (TGF-β). We also conducted bioinformatics analyses and in vivo assessments of breast cancer growth and metastasis using xenograft models. Results We found a significant association between STIM2 overexpression and metastatic breast cancer. STIM2 overexpression activated the nuclear factor of activated T cells 1 (NFAT1) and TGF-β signaling. Knockdown of STIM2 inhibited the motility of breast cancer cells by inhibiting EMT via specific suppression of NFAT1 and inhibited mammary tumor metastasis in mice. In contrast, STIM2 overexpression promoted metastasis. These findings were validated in human tissue arrays of 340 breast cancer samples for STIM2. Conclusion Taken together, our results demonstrated that STIM2 specifically regulates NFAT1, which in turn regulates the expression and secretion of TGF-β1 to promote EMT in vitro and in vivo, leading to metastasis of breast cancer. Electronic supplementary material The online version of this article (10.1186/s13058-019-1185-1) contains supplementary material, which is available to authorized users.

Online Supplementary Methods Section. (DOCX 20 kb)

IACUC Aprroval Document for Mouse Breast Cancer Xenograft Tumor Assessments. (PDF 643 kb)

IRB Approval Document for Human Breast Cancer Tissue Microarray Assessments. (PDF 2603 kb)

Phenytoin (PHT) is a commonly prescribed first-line antiepileptic drug. However, long-term administration of PHT can cause memory loss and balance disturbance. Gastrodin (GD) is the major bioactive component in Tianma and has sedative, anticonvulsive, memory strengthening, and neuroprotective effects. To combine the two drugs seems attractive; however, little was known about the efficacy of combination therapy. In this study, convulsive attack was successfully induced by penicillin. Isobolographic analysis, memory and balance behavior test, histopathological examination, and Western blot analysis were used to investigate whether the combination therapy of GD and PHT can enhance anticonvulsive effect and reduce the side effects associated with PHT. The GD alone (950.60 mg/kg) and the PHT alone (45.50 mg/kg) could produce an anticonvulsive effect, while comparable effect could be produced by PHT : GD = 1 : 50 (8.59 : 429.27 mg/kg), which reduce the dose of PHT by 81% and GD by 55%. After the chronic anticonvulsive experiments of 16 days, the balance disturbance and short-/long-term memory loss were observed in the PHT group, while the PHT + GD therapy can protect the normal balance and memory function. The neuron morphology of hippocampus was preserved, and the number of surviving neurons after combination therapy was more than the model group. The amount of NF-κB (p65) expression was increased in combination group. All above suggested the potential of the combination of PHT and GD enhances the anticonvulsive effect and the neuroprotective effect and reduces the PHT-associated memory and balance disturbance. The PHT + GD strategy would provide new possibilities as a novel promising methodology to treat epileptic patients.

Chromosome conformation capture (3C) technologies can be used to investigate 3D genomic structures. However, high background noise, high costs, and a lack of straightforward noise evaluation in current methods impede the advancement of 3D genomic research. Here we developed a simple digestion-ligation-only Hi-C (DLO Hi-C) technology to explore the 3D landscape of the genome. This method requires only two rounds of digestion and ligation, without the need for biotin labeling and pulldown. Non-ligated DNA was efficiently removed in a cost-effective step by purifying specific linker-ligated DNA fragments. Notably, random ligation could be quickly evaluated in an early quality-control step before sequencing. Moreover, an in situ version of DLO Hi-C using a four-cutter restriction enzyme has been developed. We applied DLO Hi-C to delineate the genomic architecture of THP-1 and K562 cells and uncovered chromosomal translocations. This technology may facilitate investigation of genomic organization, gene regulation, and (meta)genome assembly.

To analyze related behaviors of individual preparedness and influencing factors on violent terrorist attacks among undergraduates.A total of 1 800 undergraduates from 5 colleges or universities in Guangzhou were selected, using the stratified cluster method. A questionnaire involving the response to violent terrorist attack behavior was used to assess the individual preparedness behaviors among undergraduates. A self-made questionnaire was applied to collect information on demographic factors, cognitive and preparedness behaviors.The mean score of individual preparedness behavior among undergraduates was 13.49 ± 5.02 while information on seeking behavior was 4.27 ± 1.64, avoidance behavior was 5.97 ± 2.16 and violent terrorist attack response behaviors was 23.73 ± 7.21, with 30.0 percent of undergraduates behaved properly. Significant differences were found in the scores of behaviors on the response to violent terrorist attack with different gender, major they pursue or religious belief (P0.05), among undergraduates involved in this study. Results from the logistic regression analysis revealed that persons being girls (OR = 1.46, 95% CI: 1.06-2.01), with bigger perceived probability (OR = 1.60, 95% CI: 1.12-2.30), with higher alertness (OR = 3.77, 95% CI: 2.15-6.61), with stronger coping confidence (OR = 0.34, 95% CI: 0.24-0.48) and bigger affective response (OR1 = 3.42, 95% CI: 2.40-4.86; OR2 = 0.23, 95% CI: 0.13-0.41), would present more prominent behavior responses when facing the violent terrorist attack.Individual response behaviors to violent terrorist attacks among undergraduates were relatively ideal. Perceived probability, alertness, coping confidence and affective response appeared to be independent influencing factors related to response behaviors against violent terrorist attack. In colleges and universities, awareness on violent terrorist attacks should be strengthened among undergraduates. Focus should target on psychological education dealing with disaster, knowledge and skills needed for emergency response, so as to respond to the emergency on campus.

To explore the quality of life and related factors among people living with HIV/AIDS in Guangdong province.A total of 443 subjects older than age 18 who had been diagnosed as HIV/AIDS were recruited from AIDS Outpatient Clinic Departments in Guangzhou and Shenzhen. HIV/AIDS-targeted quality of life(HAT-QoL)scale was used to measure the quality of life on the patients. Nonparametric test method and multiple linear regression analysis were performed to explore the related factors.The average scores of the following nine domains of HIV/AIDS-targeted quality of life scales were:overall function as 59.89 ± 23.35, concerns on medication as 62.91 ± 24.01, worries on factors as being disclosed, on health, on financial situation were 19.97 ± 23.04, 54.05 ± 12.24 and 35.94 ± 26.45, respectively, trust on providers as 68.72 ± 26.17, rate of understanding HIV as 61.46 ± 25.72, on life satisfaction as 56.90 ± 25.30 and on sexual function as 50.82 ± 23.86. Factors associated with quality of life would include sex, age, residential area, education, employment, marital status, complications, HAART-related prescriptions, route of infection and duration of HIV/AIDS etc.The quality of life among HIV/AIDS patients had declined, especially on worries related to disclosure and financial situation. More attention should be paid on those with poor education, low income and lack of family support in order to improve their quality of life and reducing mortality.