Text mining for contexts and relationships in cancer genomics literature.

MOTIVATION: Scientific advances build on the findings of existing research. The 2001 publication of the human genome has led to the production of huge volumes of literature exploring the context-specific functions and interactions of genes. Technology is needed to perform large-scale text mining of research papers to extract the reported actions of genes in specific experimental contexts and cell states, such as cancer, thereby facilitating the design of new therapeutic strategies. RESULTS: We present a new corpus and Text Mining methodology that can accurately identify and extract the most important details of cancer genomics experiments from biomedical texts. We build a Named Entity Recognition model that accurately extracts relevant experiment details from PubMed abstract text, and a second model that identifies the relationships between them. This system outperforms earlier models and enables the analysis of gene function in diverse and dynamically evolving experimental contexts. AVAILABILITY AND IMPLEMENTATION: Code and data are available here: https://github.com/cambridgeltl/functional-genomics-ie.

ATX-LPA-Dependent Nuclear Translocation of Endonuclease G in Respiratory Epithelial Cells: A New Mode Action for DNA Damage Induced by Crystalline Silica Particles.

Crystalline silica particles (CSi) are an established human carcinogen, but it is not clear how these particles cause necessary mutations. A well-established scenario includes inflammation caused by retained particles in the bronchioles, activated macrophages, and reactive oxygen species (ROS) that cause DNA damage. In previous studies, we showed that CSi in contact with the plasma membrane of human bronchial epithelium induced double strand breaks within minutes. A signaling pathway implicating the ATX-LPA axis, Rac1, NLRP3, and mitochondrial depolarization upstream of DSB formation was delineated. In this paper, we provide in vitro and in vivo evidence that this signaling pathway triggers endonuclease G (EndoG) translocation from the mitochondria to the nucleus. The DNA damage is documented as γH2AX and p53BP1 nuclear foci, strand breaks in the Comet assay, and as micronuclei. In addition, the DNA damage is induced by low doses of CSi that do not induce apoptosis. By inhibiting the ATX-LPA axis or by EndoG knockdown, we prevent EndoG translocation and DSB formation. Our data indicate that CSi in low doses induces DSBs by sub-apoptotic activation of EndoG, adding CSi to a list of carcinogens that may induce mutations via sub-apoptotic and "minority MOMP" effects. This is the first report linking the ATX-LPA axis to this type of carcinogenic effect.

Application of Text Mining in Risk Assessment of Chemical Mixtures: A Case Study of Polycyclic Aromatic Hydrocarbons (PAHs).

BACKGROUND: Cancer risk assessment of complex exposures, such as exposure to mixtures of polycyclic aromatic hydrocarbons (PAHs), is challenging due to the diverse biological activities of these compounds. With the help of text mining (TM), we have developed TM tools-the latest iteration of the Cancer Risk Assessment using Biomedical literature tool (CRAB3) and a Cancer Hallmarks Analytics Tool (CHAT)-that could be useful for automatic literature analyses in cancer risk assessment and research. Although CRAB3 analyses are based on carcinogenic modes of action (MOAs) and cover almost all the key characteristics of carcinogens, CHAT evaluates literature according to the hallmarks of cancer referring to the alterations in cellular behavior that characterize the cancer cell. OBJECTIVES: The objective was to evaluate the usefulness of these tools to support cancer risk assessment by performing a case study of 22 European Union and U.S. Environmental Protection Agency priority PAHs and diesel exhaust and a case study of PAH interactions with silica. METHODS: We analyzed PubMed literature, comprising 57,498 references concerning priority PAHs and complex PAH mixtures, using CRAB3 and CHAT. RESULTS: CRAB3 analyses correctly identified similarities and differences in genotoxic and nongenotoxic MOAs of the 22 priority PAHs and grouped them according to their known carcinogenic potential. CHAT had the same capacity and complemented the CRAB output when comparing, for example, benzo[a]pyrene and dibenzo[a,l]pyrene. Both CRAB3 and CHAT analyses highlighted potentially interacting mechanisms within and across complex PAH mixtures and mechanisms of possible importance for interactions with silica. CONCLUSION: These data suggest that our TM approach can be useful in the hazard identification of PAHs and mixtures including PAHs. The tools can assist in grouping chemicals and identifying similarities and differences in carcinogenic MOAs and their interactions. https://doi.org/10.1289/EHP6702.

Crystalline silica particles induce DNA damage in respiratory epithelium by ATX secretion and Rac1 activation.

Autotaxin (ATX) and its product lysophosphatidic acid (LPA) have been implicated in lung fibrosis and cancer. We have studied their roles in DNA damage induced by carcinogenic crystalline silica particles (CSi). In an earlier study on bronchial epithelia, we concluded that ATX, via paracrine signaling, amplifies DNA damage. This effect was seen at 6-16 h. A succeeding study showed that CSi induced NLRP3 phosphorylation, mitochondrial depolarization, double strand breaks (DSBs), and NHEJ repair enzymes within minutes. In the current study we hypothesized a role for the ATX-LPA axis also in this rapid DNA damage. Using 16HBE human bronchial epithelial cells, we show ATX secretion at 3 min, and that ATX inhibitors (HA130 and PF8380) prevented both CSi-induced mitochondrial depolarization and DNA damage (detected by γH2AX and Comet assay analysis). Experiments with added LPA gave similar rapid effects as CSi. Furthermore, Rac1 was activated at 3 min, and a Rac1 inhibitor (NSC23766) prevented mitochondrial depolarization and genotoxicity. In mice the bronchial epithelia exhibited histological signs of ATX activation and signs of DSBs (53BP1 positive nuclei) minutes after a single inhalation of CSi. Our data indicate that CSi rapidly activate the ATX-LPA axis and within minutes this leads to DNA damage in bronchial epithelial cells. Thus, ATX mediates very rapid DNA damaging effects of inhaled particles.

Crystalline silica particles cause rapid NLRP3-dependent mitochondrial depolarization and DNA damage in airway epithelial cells.

BACKGROUND: Respirable crystalline silica causes lung carcinomas and many thousand future cancer cases are expected in e.g. Europe. Critical questions are how silica causes genotoxicity in the respiratory epithelium and if new cases can be avoided by lowered permissible exposure levels. In this study we investigate early DNA damaging effects of low doses of silica particles in respiratory epithelial cells in vitro and in vivo in an effort to understand low-dose carcinogenic effects of silica particles. RESULTS: We find DNA damage accumulation already after 5-10 min exposure to low doses (5 µg/cm2) of silica particles (Min-U-Sil 5) in vitro. DNA damage was documented as increased levels of γH2AX, pCHK2, by Comet assay, AIM2 induction, and by increased DNA repair (non-homologous end joining) signaling. The DNA damage response (DDR) was not related to increased ROS levels, but to a NLRP3-dependent mitochondrial depolarization. Particles in contact with the plasma membrane elicited a Ser198 phosphorylation of NLRP3, co-localization of NLRP3 to mitochondria and depolarization. FCCP, a mitochondrial uncoupler, as well as overexpressed NLRP3 mimicked the silica-induced depolarization and the DNA damage response. A single inhalation of 25 µg silica particles gave a similar rapid DDR in mouse lung. Biomarkers (CC10 and GPRC5A) indicated an involvement of respiratory epithelial cells. CONCLUSIONS: Our findings demonstrate a novel mode of action (MOA) for silica-induced DNA damage and mutagenic double strand breaks in airway epithelial cells. This MOA seems independent of particle uptake and of an involvement of macrophages. Our study might help defining models for estimating exposure levels without DNA damaging effects.

PTEN and PHLPP crosstalk in cancer cells and in TGFß-activated stem cells.

Akt kinase regulates several cellular processes, among them growth, proliferation and survival, and has been correlated to neoplastic disease. We report here crosstalk between several Akt regulatory phosphatases that controls the level of the activated form (phosphorylated) of Akt and affects tumor cell aggressiveness. In prostate cancer cell lines, we observed that transient transfection of PTEN decreased the endogenous level of PHLPPs and in contrast, the transient transfection of PHLPPs decreased the endogenous level of PTEN. Furthermore, silencing of PTEN by siRNA resulted in increased PHLPP levels. This phenomenon was not seen in non-transformed cells or in prostate stem cells. This crosstalk promoted cancer cell invasion and was controlled by epigenetically regulated processes where activation of miRs (miR-190 and miR214), the polycomb group of proteins and DNA methylation were involved. The purinergic P2X4 receptor, which has been shown to have a role in wound healing, was identified to be the mediator of this crosstalk. We also studied prostate stem cells and found this crosstalk in the TGFß1-activated epithelial-mesenchymal transition (EMT). The crosstalk seemed to be a natural part of EMT. In summary, we identify a crosstalk between Akt phosphatases which is not present in non-transformed prostate cells but occurs in cancer cells and stem cells transformed by TGFß-1. This crosstalk is important for cellular invasion. BACKGROUND: Phosphatases regulate the Akt oncogene. RESULTS: Crosstalk between Akt phosphatases in prostate cancer cells and in TGF-ß1 activated stem cells but not in non-transformed cells. CONCLUSION: This back-up mechanism facilitates invasive migration of prostate stem and cancer cells. SIGNIFICANCE: Characterization of Akt regulation may lead to a better understanding of tumor development and to novel strategies for treatment.

Pesticide exposure among Bolivian farmers: associations between worker protection and exposure biomarkers.

The use of pesticides has increased during the past decades, also increasing the risk of exposure to toxic pesticides that can cause detrimental health effects in the future. This is of special concern among farmers in low-to-middle-income countries that may lack proper training in the safe use of these chemicals. To assess the situation in Bolivia a cross-sectional study in three agricultural communities was performed (n = 297). Handling, use of personal protective equipment (PPE) and pesticide exposure were assessed by a questionnaire and measurements of urinary pesticide metabolites (UPMs). Results showed that methamidophos (65%) and paraquat (52%) were the most commonly used pesticides and that 75% of the farmers combined several pesticides while spraying. Notably, only 17% of the farmers used recommended PPEs while 84% reported to have experienced symptoms of acute pesticide poisoning after spraying. UPM measurements indicated high levels of exposure to chlorpyrifos, pyrethroids and 2,4D and that men generally were more highly exposed compared to women. Our study demonstrates that farmers who are better at following recommendations for pesticide handling and use of PPE had a significantly lower risk of having high UPM levels of most measured pesticides. Our results thus confirm the need of proper training of farmers in low-to-middle-income countries in proper protection and pesticide handling in order to reduce exposure levels and health problems.

Correction to: Pesticide exposure among Bolivian farmers: associations between worker protection and exposure biomarkers.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.In the original Article, co-author Ulla Stenius' surname was misspelled as Ulla Steinus. This has been corrected in the PDF, HTML and XML versions of this Article.

Increased levels of genotoxic damage in a Bolivian agricultural population exposed to mixtures of pesticides.

During the past decades, farmers in low to middle-income countries have increased their use of pesticides, and thereby the risk of being exposed to potentially genotoxic chemicals that can cause adverse health effects. Here, the aim was to investigate the correlation between exposure to pesticides and genotoxic damage in a Bolivian agricultural population. Genotoxic effects were assessed in peripheral blood samples by comet and micronucleus (MN) assays, and exposure levels by measurements of 10 urinary pesticide metabolites. Genetic susceptibility was assessed by determination of null frequency of GSTM1 and GSTT1 genotypes. The results showed higher MN frequency in women and farmers active ≥8 years compared to their counterpart (P < 0.05). In addition, age, GST genotype, alcohol consumption, and type of water source influenced levels of genotoxic damage. Individuals with high exposure to tebuconazole, 2,4-D, or cyfluthrin displayed increased levels of genotoxic damage (P < 0.05-0.001). Logistic regression was conducted to evaluate associations between pesticide exposure and risk of genotoxic damage. After adjustment for confounders, a significant increased risk of DNA strand breaks was found for high exposure to 2,4-D, odds ratio (OR) = 1.99 (P < 0.05). In contrast, high exposure to pyrethroids was associated with a reduced risk of DNA strand breaks, OR = 0.49 (P < 0.05). It was also found that high exposure to certain mixtures of pesticides (containing mainly 2,4-D or cyfluthrin) was significantly associated with increased level and risk of genotoxic damage (P < 0.05). In conclusion, our data show that high exposure levels to some pesticides is associated with an increased risk of genotoxic damage among Bolivian farmers, suggesting that their use should be better controlled or limited.

The Water Extract of Juniperus communis L. Induces Cell Death and Sensitizes Cancer Cells to Cytostatic Drugs through p53 and PI3K/Akt Pathways.

Juniper (Juniperus communis L.) is a northern coniferous plant generally used as a spice and for nutritional purposes in foods and drinks. It was previously reported that juniper extract (JE) affects p53 activity, cellular stress, and gene expression induced cell death in human neuroblastoma cells. Therefore, the effects of juniper on p53 and Akt signaling was examined further in A549 lung, 22RV1 and DU145 prostate, and HepG2 liver cancer cells using Western blot, confocal microscopy, and MTT analysis. We found that juniper simultaneously decreased cell viability, activated the p53 pathway, and inactivated the PI3K/Akt pathway. The p53 activation was associated with increased nuclear p53 level. Akt was dephosphorylated, and its inactivation was associated with increased levels of PHLPP1 and PHLPP2 phosphatases. Parallel increases of PARP suggest that JE decreased cell viability by activating cell death. In adtion, JE potentiated the effects of gemcitabine and 5-fluorouracil anticancer drugs. Thus, JE can activate cell death in different cancer cell lines through p53 and Akt pathways.

A neural classification method for supporting the creation of BioVerbNet.

BACKGROUND: VerbNet, an extensive computational verb lexicon for English, has proved useful for supporting a wide range of Natural Language Processing tasks requiring information about the behaviour and meaning of verbs. Biomedical text processing and mining could benefit from a similar resource. We take the first step towards the development of BioVerbNet: A VerbNet specifically aimed at describing verbs in the area of biomedicine. Because VerbNet-style classification is extremely time consuming, we start from a small manual classification of biomedical verbs and apply a state-of-the-art neural representation model, specifically developed for class-based optimization, to expand the classification with new verbs, using all the PubMed abstracts and the full articles in the PubMed Central Open Access subset as data. RESULTS: Direct evaluation of the resulting classification against BioSimVerb (verb similarity judgement data in biomedicine) shows promising results when representation learning is performed using verb class-based contexts. Human validation by linguists and biologists reveals that the automatically expanded classification is highly accurate. Including novel, valid member verbs and classes, our method can be used to facilitate cost-effective development of BioVerbNet. CONCLUSION: This work constitutes the first effort on applying a state-of-the-art architecture for neural representation learning to biomedical verb classification. While we discuss future optimization of the method, our promising results suggest that the automatic classification released with this article can be used to readily support application tasks in biomedicine.

TGF beta promotes repair of bulky DNA damage through increased ERCC1/XPF and ERCC1/XPA interaction.

Transforming growth factor beta (TGFß) is multifunctional cytokine that is involved in the coordination and regulation of many cellular homeostatic processes. Compromised TGFß activity has been attributed to promotion of human cancers. Recent studies have identified a role for TGFß in response to radiation-induced DNA damage, suggesting a link between TGFß and the DNA damage response with implications for cancer development. In this study, the effects of TGFß on promoting the repair of bulky DNA damage, through modulation of nucleotide excision repair (NER), were investigated. We show that treatment of cells with exogenous TGFß leads to enhanced repair of DNA damage formed by polycyclic aromatic hydrocarbons and ultraviolet-C radiation; similarly, cells with constitutively activated endogenous TGFß signaling show comparable responses. This effect of TGFß is independent of the cell cycle. The response to TGFß is decreased in cells that have compromised TGFß signaling through RNA interference of Smad4 and is decreased in NER-deficient cells and cells with compromised NER through RNA interference of excision repair cross-complementing group 1 (ERCC1). Increased interaction and nuclear localization of ERCC1/xeroderma pigmentosum (XP) F and ERCC1/XPA proteins is observed after TGFß treatment. Our study represents the first experimental evidence of a role for TGFß in the repair of bulky DNA damage resulting from promotion of the interaction and localization of repair protein complexes involved in the incision step of NER.

CXADR-Mediated Formation of an AKT Inhibitory Signalosome at Tight Junctions Controls Epithelial-Mesenchymal Plasticity in Breast Cancer.

Tight junctions (TJ) act as hubs for intracellular signaling pathways controlling epithelial cell fate and function. Deregulation of TJ is a hallmark of epithelial-mesenchymal transition (EMT), which contributes to carcinoma progression and metastasis. However, the signaling mechanisms linking TJ to the induction of EMT are not understood. Here, we identify a TJ-based signalosome, which controls AKT signaling and EMT in breast cancer. The coxsackie and adenovirus receptor (CXADR), a TJ protein with an essential yet uncharacterized role in organogenesis and tissue homeostasis, was identified as a key component of the signalosome. CXADR regulated the stability and function of the phosphatases and AKT inhibitors PTEN and PHLPP2. Loss of CXADR led to hyperactivation of AKT and sensitized cells to TGFß1-induced EMT. Conversely, restoration of CXADR stabilized PHLPP2 and PTEN, inhibited AKT, and promoted epithelial differentiation. Loss of CXADR in luminal A breast cancer correlated with loss of PHLPP2 and PTEN and poor prognosis. These results show that CXADR promotes the formation of an AKT-inhibitory signalosome at TJ and regulates epithelial-mesenchymal plasticity in breast cancer cells. Moreover, loss of CXADR might be used as a prognostic marker in luminal breast cancer. SIGNIFICANCE: The tight junction protein CXADR controls epithelial-mesenchymal plasticity in breast cancer by stabilizing the AKT regulators PTEN and PHLPP2.Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/79/1/47/F1.large.jpg.

LION LBD: a literature-based discovery system for cancer biology.

MOTIVATION: The overwhelming size and rapid growth of the biomedical literature make it impossible for scientists to read all studies related to their work, potentially leading to missed connections and wasted time and resources. Literature-based discovery (LBD) aims to alleviate these issues by identifying implicit links between disjoint parts of the literature. While LBD has been studied in depth since its introduction three decades ago, there has been limited work making use of recent advances in biomedical text processing methods in LBD. RESULTS: We present LION LBD, a literature-based discovery system that enables researchers to navigate published information and supports hypothesis generation and testing. The system is built with a particular focus on the molecular biology of cancer using state-of-the-art machine learning and natural language processing methods, including named entity recognition and grounding to domain ontologies covering a wide range of entity types and a novel approach to detecting references to the hallmarks of cancer in text. LION LBD implements a broad selection of co-occurrence based metrics for analyzing the strength of entity associations, and its design allows real-time search to discover indirect associations between entities in a database of tens of millions of publications while preserving the ability of users to explore each mention in its original context in the literature. Evaluations of the system demonstrate its ability to identify undiscovered links and rank relevant concepts highly among potential connections. AVAILABILITY AND IMPLEMENTATION: The LION LBD system is available via a web-based user interface and a programmable API, and all components of the system are made available under open licenses from the project home page http://lbd.lionproject.net. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Toluene diisocyanate exposure and autotaxin-lysophosphatidic acid signalling.

Toluene diisocyanate (TDI) is a reactive chemical used in manufacturing plastics. TDI exposure adversely affects workers' health, causing occupational asthma, but individuals differ in susceptibility. We recently suggested a role for signalling mediated by the enzyme autotaxin (ATX) and its product, lysophosphatidic acid (LPA), in TDI toxicity. Here we genotyped 118 TDI-exposed workers for six single-nucleotide polymorphisms (SNPs) in genes encoding proteins implicated in ATX-LPA signalling: purinergic receptor P2X7 (P2RX7), CC motif chemokine ligand 2 (CCL2), interleukin 1ß (IL1B), and caveolin 1 (CAV1). Two P2RX7 SNPs (rs208294 and rs2230911) significantly modified the associations between a biomarker of TDI exposure (urinary 2,4-toluene diamine) and plasma LPA; two IL1B SNPs (rs16944 and rs1143634) did not. CAV1 rs3807989 modified the associations, but the effect was not statistically significant (p = 0.05-0.09). In vitro, TDI-exposed bronchial epithelial cells (16HBE14o-) rapidly released ATX and IL-1ß. P2X7 inhibitors attenuated both responses, but confocal microscopy showed non-overlapping localizations of ATX and IL-1ß, and down-regulation of CAV1 inhibited the ATX response but not the IL-1ß response. This study indicates that P2X7 is pivotal for TDI-induced ATX-LPA signalling, which was modified by genetic variation in P2RX7. Furthermore, our data suggest that the TDI-induced ATX and IL-1ß responses occur independently.

Cancer Hallmarks Analytics Tool (CHAT): a text mining approach to organize and evaluate scientific literature on cancer.

MOTIVATION: To understand the molecular mechanisms involved in cancer development, significant efforts are being invested in cancer research. This has resulted in millions of scientific articles. An efficient and thorough review of the existing literature is crucially important to drive new research. This time-demanding task can be supported by emerging computational approaches based on text mining which offer a great opportunity to organize and retrieve the desired information efficiently from sizable databases. One way to organize existing knowledge on cancer is to utilize the widely accepted framework of the Hallmarks of Cancer. These hallmarks refer to the alterations in cell behaviour that characterize the cancer cell. RESULTS: We created an extensive Hallmarks of Cancer taxonomy and developed automatic text mining methodology and a tool (CHAT) capable of retrieving and organizing millions of cancer-related references from PubMed into the taxonomy. The efficiency and accuracy of the tool was evaluated intrinsically as well as extrinsically by case studies. The correlations identified by the tool show that it offers a great potential to organize and correctly classify cancer-related literature. Furthermore, the tool can be useful, for example, in identifying hallmarks associated with extrinsic factors, biomarkers and therapeutics targets. AVAILABILITY AND IMPLEMENTATION: CHAT can be accessed at: http://chat.lionproject.net. The corpus of hallmark-annotated PubMed abstracts and the software are available at: http://chat.lionproject.net/about. CONTACT: simon.baker@cl.cam.ac.uk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

ATM-activated autotaxin (ATX) propagates inflammation and DNA damage in lung epithelial cells: a new mode of action for silica-induced DNA damage?

Silica exposure is a common risk factor for lung cancer. It has been claimed that key elements in cancer development are activation of inflammatory cells that indirectly induce DNA damage and proliferative stimuli in respiratory epithelial cells. We studied DNA damage induced by silica particles in respiratory epithelial cells and focused the role of the signaling enzyme autotaxin (ATX). A549 and 16 bronchial epithelial cells (16HBE) lung epithelial cells were exposed to silica particles. Reactive oxygen species (ROS), NOD-like receptor family pyrin domain containing-3 (NLRP3) inflammasome activation, ATX, ataxia telangiectasia mutated (ATM), and DNA damage (γH2AX, pCHK1, pCHK2, comet assay) were end points. Low doses of silica induced NLRP3 activation, DNA damage accumulation, and ATM phosphorylation. A novel finding was that ATM induced ATX generation and secretion. Not only silica but also rotenone, camptothecin and H2O2 activated ATX via ATM, suggesting that ATX is part of a generalized ATM response to double-strand breaks (DSBs). Surprisingly, ATX inhibition mitigated DNA damage accumulation at later time points (6-16 h), and ATX transfection caused NLRP3 activation and DNA damage. Furthermore, the product of ATX enzymatic activity, lysophosphatidic acid, recapitulated the effects of ATX transfection. These data indicate an ATM-ATX-dependent loop that propagates inflammation and DSB accumulation, making low doses of silica effective inducers of DSBs in epithelial cells. We conclude that an ATM-ATX axis interconnects DSBs with silica-induced inflammation and propagates these effects in epithelial cells. Further studies of this adverse outcome pathway may give an accurate assessment of the lowest doses of silica that causes cancer.

Cancer Risk Assessment of Airborne PAHs Based on in Vitro Mixture Potency Factors.

Complex mixtures of polycyclic aromatic hydrocarbons (PAHs) are common environmental pollutants associated with adverse human health effects including cancer. However, the risk of exposure to mixtures is difficult to estimate, and risk assessment by whole mixture potency evaluations has been suggested. To facilitate this, reliable in vitro based testing systems are necessary. Here, we investigated if activation of DNA damage signaling in vitro could be an endpoint for developing whole mixture potency factors (MPFs) for airborne PAHs. Activation of DNA damage signaling was assessed by phosphorylation of Chk1 and H2AX using Western blotting. To validate the in vitro approach, potency factors were determined for seven individual PAHs which were in very good agreement with established potency factors based on cancer data in vivo. Applying the method using Stockholm air PAH samples indicated MPFs with orders of magnitude higher carcinogenic potency than predicted by established in vivo-based potency factors. Applying the MPFs in cancer risk assessment suggested that 45.4 (6% of all) cancer cases per year in Stockholm are due to airborne PAHs. Applying established models resulted in <1 cancer case per year, which is far from expected levels. We conclude that our in vitro based approach for establishing MPFs could be a novel method to assess whole mixture samples of airborne PAHs to improve health risk assessment.

Text mining for improved exposure assessment.

Chemical exposure assessments are based on information collected via different methods, such as biomonitoring, personal monitoring, environmental monitoring and questionnaires. The vast amount of chemical-specific exposure information available from web-based databases, such as PubMed, is undoubtedly a great asset to the scientific community. However, manual retrieval of relevant published information is an extremely time consuming task and overviewing the data is nearly impossible. Here, we present the development of an automatic classifier for chemical exposure information. First, nearly 3700 abstracts were manually annotated by an expert in exposure sciences according to a taxonomy exclusively created for exposure information. Natural Language Processing (NLP) techniques were used to extract semantic and syntactic features relevant to chemical exposure text. Using these features, we trained a supervised machine learning algorithm to automatically classify PubMed abstracts according to the exposure taxonomy. The resulting classifier demonstrates good performance in the intrinsic evaluation. We also show that the classifier improves information retrieval of chemical exposure data compared to keyword-based PubMed searches. Case studies demonstrate that the classifier can be used to assist researchers by facilitating information retrieval and classification, enabling data gap recognition and overviewing available scientific literature using chemical-specific publication profiles. Finally, we identify challenges to be addressed in future development of the system.