A framework for structured semantic representation capable of active sensing and interpretable inference: A cancer prognostic analysis case study.

Precise semantic representation is important for allowing machines to truly comprehend the meaning of natural language text, especially biomedical literature. Although the semantic relations among words in a single sentence may be accurately represented with existing approaches, relations between two sentences cannot yet be accurately modeled, which leads to a lack of contextual information and difficulty in performing interpretable semantic inference. Additionally, it is challenging to merge semantic representations curated by different experts. These critical challenges are insufficiently addressed by existing methods. In this paper, we present a framework for structured semantic representation (FSSR) to address these issues. FSSR uses a double-layer structure Construct that combines Paradigm and Instance to represent the semantics of a word or a sentence. It uses six types of rules to represent the semantic relations between sentence Constructs and uses a Computational Model to represent an action. FSSR is a graph-based representation of semantics, in which a node represents a Construct or a Paradigm. Two nodes are connected by an edge (a rule). In addition, FSSR enables interpretable inference and active acquisition of new information, as illustrated in a case study. This case study models the semantics of a cancer prognostic analysis article and reproduces its text results and charts. We provide a website that visualizes the inference process (http://cragraph.synergylab.cn).

Predicted mouse interactome and network-based interpretation of differentially expressed genes.

The house mouse or Mus musculus has become a premier mammalian model for genetic research due to its genetic and physiological similarities to humans. It brought mechanistic insights into numerous human diseases and has been routinely used to assess drug efficiency and toxicity, as well as to predict patient responses. To facilitate molecular mechanism studies in mouse, we present the Mouse Interactome Database (MID, Version 1), which includes 155,887 putative functional associations between mouse protein-coding genes inferred from functional association evidence integrated from 9 public databases. These putative functional associations are expected to cover 19.32% of all mouse protein interactions, and 26.02% of these function associations may represent protein interactions. On top of MID, we developed a gene set linkage analysis (GSLA) web tool to annotate potential functional impacts from observed differentially expressed genes. Two case studies show that the MID/GSLA system provided precise and informative annotations that other widely used gene set annotation tools, such as PANTHER and DAVID, did not. Both MID and GSLA are accessible through the website http://mouse.biomedtzc.cn.

HIR V2: a human interactome resource for the biological interpretation of differentially expressed genes via gene set linkage analysis.

To facilitate biomedical studies of disease mechanisms, a high-quality interactome that connects functionally related genes is needed to help investigators formulate pathway hypotheses and to interpret the biological logic of a phenotype at the biological process level. Interactions in the updated version of the human interactome resource (HIR V2) were inferred from 36 mathematical characterizations of six types of data that suggest functional associations between genes. This update of the HIR consists of 88 069 pairs of genes (23.2% functional interactions of HIR V2 are in common with the previous version of HIR), representing functional associations that are of strengths similar to those between well-studied protein interactions. Among these functional interactions, 57% may represent protein interactions, which are expected to cover 32% of the true human protein interactome. The gene set linkage analysis (GSLA) tool is developed based on the high-quality HIR V2 to identify the potential functional impacts of the observed transcriptomic changes, helping to elucidate their biological significance and complementing the currently widely used enrichment-based gene set interpretation tools. A case study shows that the annotations reported by the HIR V2/GSLA system are more comprehensive and concise compared to those obtained by the widely used gene set annotation tools such as PANTHER and DAVID. The HIR V2 and GSLA are available at http://human.biomedtzc.cn.

Predicted rat interactome database and gene set linkage analysis.

Rattus norvegicus, or the rat, has been widely used as animal models for a diversity of human diseases in the last 150 years. The rat, as a disease model, has the advantage of relatively large body size and highly similar physiology to humans. In drug discovery, rat models are routinely used in drug efficacy and toxicity assessments. To facilitate molecular pharmacology studies in rats, we present the predicted rat interactome database (PRID), which is a database of high-quality predicted functional gene interactions with balanced sensitivity and specificity. PRID integrates functional gene association data from 10 public databases and infers 305 939 putative functional associations, which are expected to include 13.02% of all rat protein interactions, and 52.59% of these function associations may represent protein interactions. This set of functional interactions may not only facilitate hypothesis formulation in molecular mechanism studies, but also serve as a reference interactome for users to perform gene set linkage analysis (GSLA), which is a web-based tool to infer the potential functional impacts of a set of changed genes observed in transcriptomics analyses. In a case study, we show that GSLA based on PRID may provide more precise and informative annotations for investigators to understand the physiological mechanisms underlying a phenotype and lead investigators to testable hypotheses for further studies. Widely used functional annotation tools such as Gene Ontology (GO) analysis, and Database for Annotation, Visualization and Integrated Discovery (DAVID) did not provide similar insights. Database URL: http://rat.biomedtzc.cn.

Predicted functional interactome of Caenorhabditis elegans and a web tool for the functional interpretation of differentially expressed genes.

BACKGROUND: The nematode worm, Caenorhabditis elegans, is a saprophytic species that has been emerging as a standard model organism since the early 1960s. This species is useful in numerous fields, including developmental biology, neurobiology, and ageing. A high-quality comprehensive molecular interaction network is needed to facilitate molecular mechanism studies in C. elegans. RESULTS: We present the predicted functional interactome of Caenorhabditis elegans (FIC), which integrates functional association data from 10 public databases to infer functional gene interactions on diverse functional perspectives. In this work, FIC includes 108,550 putative functional associations with balanced sensitivity and specificity, which are expected to cover 21.42% of all C. elegans protein interactions, and 29.25% of these associations may represent protein interactions. Based on FIC, we developed a gene set linkage analysis (GSLA) web tool to interpret potential functional impacts from a set of differentially expressed genes observed in transcriptome analyses. CONCLUSION: We present the predicted C. elegans interactome database FIC, which is a high-quality database of predicted functional interactions among genes. The functional interactions in FIC serve as a good reference interactome for GSLA to annotate differentially expressed genes for their potential functional impacts. In a case study, the FIC/GSLA system shows more comprehensive and concise annotations compared to other widely used gene set annotation tools, including PANTHER and DAVID. FIC and its associated GSLA are available at the website http://worm.biomedtzc.cn .

Predicted yeast interactome and network-based interpretation of transcriptionally changed genes.

Saccharomyces cerevisiae, budding yeast, is a widely used model organism and research tool in genetics studies. Many efforts have been directed at constructing a high-quality comprehensive molecular interaction network to elucidate the design logic of the gene circuitries in this classic model organism. In this work, we present the yeast interactome resource (YIR), which includes 22,238 putative functional gene interactions inferred from functional gene association data integrated from 10 databases focusing on diverse functional perspectives. These putative functional gene interactions are expected to cover 18.84% of yeast protein interactions, and 38.49% may represent protein interactions. Based on the YIR, a gene set linkage analysis (GSLA) web tool was developed to annotate the potential functional impacts of a set of transcriptionally changed genes. In a case study, we show that the YIR/GSLA system produced more extensive and concise annotations compared with widely used gene set annotation tools, including PANTHER and DAVID. Both YIR and GSLA are accessible through the website http://yeast.biomedtzc.cn.

Predicted Drosophila Interactome Resource and web tool for functional interpretation of differentially expressed genes.

Drosophila melanogaster is a well-established model organism that is widely used in genetic studies. This species enjoys the availability of a wide range of research tools, well-annotated reference databases and highly similar gene circuitry to other insects. To facilitate molecular mechanism studies in Drosophila, we present the Predicted Drosophila Interactome Resource (PDIR), a database of high-quality predicted functional gene interactions. These interactions were inferred from evidence in 10 public databases providing information for functional gene interactions from diverse perspectives. The current version of PDIR includes 102 835 putative functional associations with balanced sensitivity and specificity, which are expected to cover 22.56% of all Drosophila protein interactions. This set of functional interactions is a good reference for hypothesis formulation in molecular mechanism studies. At the same time, these interactions also serve as a high-quality reference interactome for gene set linkage analysis (GSLA), which is a web tool for the interpretation of the potential functional impacts of a set of changed genes observed in transcriptomics analyses. In a case study, we show that the PDIR/GSLA system was able to produce a more comprehensive and concise interpretation of the collective functional impact of multiple simultaneously changed genes compared with the widely used gene set annotation tools, including PANTHER and David. PDIR and its associated GSLA service can be accessed at http://drosophila.biomedtzc.cn.

Developmental neurotoxicity and immunotoxicity induced by graphene oxide in zebrafish embryos.

Graphene oxide (GO) has emerged as the worldwide promising candidate for biomedical application, such as for drug delivery, bio-sensing and anti-cancer therapy. This study was focused on the zebrafish and RAW264.7 cell line as in vivo and in vitro models to assess the potential developmental neurotoxicity and immunotoxicity of GO. No obvious acute developmental toxicity was observed upon treatments with 0.01, 0.1, and 1 µg/mL GO for five consecutive days. However, decreased hatching rate, increased malformation rate, heart beat rate and hypoactivity of locomotor behavior were detected when exposed to 10 µg/mL GO. Also, RT-PCR analysis revealed that expressions of genes related to the nervous system were up-regulated. The potential risk of GO for developmental neurotoxicity may be ascribed to the high level of oxidative stress induced by high concentration of GO. Most importantly, the mRNA levels of immune response associated genes, such as interleukin-6 (IL-6), interleukin-8 (IL-8), tumor necrosis factor-α (TNFα), interferon-γ (IFN-γ) were significantly increased under environmental concentration exposure. The activation of pro-inflammatory immune response was also observed in macrophage cell line. Taken together, our results demonstrated that immunotoxicity is a sensitive indicator for assessment of bio-compatibility of GO.

Titanium dioxide nanoparticle stimulating pro-inflammatory responses in vitro and in vivo for inhibited cancer metastasis.

AIMS: The interaction of engineered nanoparticles (NPs) with the immune system and the possibility of inflammation induction are of particularly interest. Titanium dioxide nanoparticles (TiO2 NPs) are one of the most popular manufactured nanomaterials. In this study, we focused on the immune-modulatory effect of commercial P-25 TiO2 NPs in vivo and in vitro and their crucial role in cancer metastasis. MAIN METHODS: The female C57BL/6 mice were injected into abdominal cavity with PBS or P-25 TiO2 to investigate the immune-modulatory function of P-25. And breast cancer cells were intravenously (i.v.) injected into mouse to establish the liver and lung cancer metastasis model. Peritoneal macrophage was used to investigate the macrophage polarization in vitro. KEY FINDINGS: Results showed us that peritoneal macrophage exposed to P-25 TiO2 NPs displayed activated M1 macrophage response, as evidenced by the increased mRNA expression of interleukin-1ß (IL1ß), IL6, TNFα, CCR7 and inducible nitric oxide synthase (iNOS). After exposure of TiO2 NPs in vivo for 21 days, the body weights of mice decreased significantly, which were accompanied by an infiltration of immune cells in liver and spleen in 20 mg/kg BW treated group. Importantly, the production of pro-inflammatory cytokines in liver, spleen and the serum were amplified, which indicated the tissue and systemic inflammation induced by TiO2 NPs. In addition, the activation of immune response induced by P-25 TiO2 NPs was correlated with their ability to inhibit cancer metastasis. SIGNIFICANCE: Our results delineated the stimulating pro-inflammatory response induced by P-25 TiO2 NPs and their outcome in vivo for cancer metastasis.