A CHCHD6-APP axis connects amyloid and mitochondrial pathology in Alzheimer's disease.

The mechanistic relationship between amyloid-beta precursor protein (APP) processing and mitochondrial dysfunction in Alzheimer's disease (AD) has long eluded the field. Here, we report that coiled-coil-helix-coiled-coil-helix domain containing 6 (CHCHD6), a core protein of the mammalian mitochondrial contact site and cristae organizing system, mechanistically connects these AD features through a circular feedback loop that lowers CHCHD6 and raises APP processing. In cellular and animal AD models and human AD brains, the APP intracellular domain fragment inhibits CHCHD6 transcription by binding its promoter. CHCHD6 and APP bind and stabilize one another. Reduced CHCHD6 enhances APP accumulation on mitochondria-associated ER membranes and accelerates APP processing, and induces mitochondrial dysfunction and neuronal cholesterol accumulation, promoting amyloid pathology. Compensation for CHCHD6 loss in an AD mouse model reduces AD-associated neuropathology and cognitive impairment. Thus, CHCHD6 connects APP processing and mitochondrial dysfunction in AD. This provides a potential new therapeutic target for patients.

Drug repurposing for opioid use disorders: integration of computational prediction, clinical corroboration, and mechanism of action analyses.

Morbidity and mortality from opioid use disorders (OUD) and other substance use disorders (SUD) is a major public health crisis, yet there are few medications to treat them. There is an urgency to accelerate SUD medication development. We present an integrated drug repurposing strategy that combines computational prediction, clinical corroboration using electronic health records (EHRs) of over 72.9 million patients and mechanisms of action analysis. Among top-ranked repurposed candidate drugs, tramadol, olanzapine, mirtazapine, bupropion, and atomoxetine were associated with increased odds of OUD remission (adjusted odds ratio: 1.51 [1.38-1.66], 1.90 [1.66-2.18], 1.38 [1.31-1.46], 1.37 [1.29-1.46], 1.48 [1.25-1.76], p value < 0.001, respectively). Genetic and functional analyses showed these five candidate drugs directly target multiple OUD-associated genes including BDNF, CYP2D6, OPRD1, OPRK1, OPRM1, HTR1B, POMC, SLC6A4 and OUD-associated pathways, including opioid signaling, G-protein activation, serotonin receptors, and GPCR signaling. In summary, we developed an integrated drug repurposing approach and identified five repurposed candidate drugs that might be of value for treating OUD patients, including those suffering from comorbid conditions.

Prediction and evaluation of combination pharmacotherapy using natural language processing, machine learning and patient electronic health records.

Combination pharmacotherapy targets key disease pathways in a synergistic or additive manner and has high potential in treating complex diseases. Computational methods have been developed to identifying combination pharmacotherapy by analyzing large amounts of biomedical data. Existing computational approaches are often underpowered due to their reliance on our limited understanding of disease mechanisms. On the other hand, observable phenotypic inter-relationships among thousands of diseases often reflect their underlying shared genetic and molecular underpinnings, therefore can offer unique opportunities to design computational models to discover novel combinational therapies by automatically transferring knowledge among phenotypically related diseases. We developed a novel phenome-driven drug discovery system, named TuSDC, which leverages knowledge of existing drug combinations, disease comorbidities, and disease treatments of thousands of disease and drug entities extracted from over 31.5 million biomedical research articles using natural language processing techniques. TuSDC predicts combination pharmacotherapy by extracting representations of diseases and drugs using tensor factorization approaches. In external validation, TuSDC achieved an average precision of 0.77 for top ranked candidates, outperforming a state of art mechanism-based method for discovering drug combinations in treating hypertension. We evaluated top ranked anti-hypertension drug combinations using electronic health records of 84.7 million unique patients and showed that a novel drug combination hydrochlorothiazide-digoxin was associated with significantly lower hazards of subsequent hypertension as compared to the monotherapy hydrochlorothiazide alone (HR: 0.769, 95% CI [0.732, 0.807]) and digoxin alone (0.857, 95% CI [0.785, 0.936]). Data-driven informatics analyses reveal that the renin-angiotensin system is involved in the synergistical interactions of hydrochlorothiazide and digoxin on regulating hypertension. The prediction model's code with PyTorch version 1.5 is available at http://nlp.case.edu/public/data/TuSDC/.

COVID-19 and dementia: Analyses of risk, disparity, and outcomes from electronic health records in the US.

INTRODUCTION: At present, there is limited data on the risks, disparity, and outcomes for COVID-19 in patients with dementia in the United States. METHODS: This is a retrospective case-control analysis of patient electronic health records (EHRs) of 61.9 million adult and senior patients (age ≥ 18 years) in the United States up to August 21, 2020. RESULTS: Patients with dementia were at increased risk for COVID-19 compared to patients without dementia (adjusted odds ratio [AOR]: 2.00 [95% confidence interval (CI), 1.94-2.06], P < .001), with the strongest effect for vascular dementia (AOR: 3.17 [95% CI, 2.97-3.37], P < .001), followed by presenile dementia (AOR: 2.62 [95% CI, 2.28-3.00], P < .001), Alzheimer's disease (AOR: 1.86 [95% CI, 1.77-1.96], P < .001), senile dementia (AOR: 1.99 [95% CI, 1.86-2.13], P < .001) and post-traumatic dementia (AOR: 1.67 [95% CI, 1.51-1.86] P < .001). Blacks with dementia had higher risk of COVID-19 than Whites (AOR: 2.86 [95% CI, 2.67-3.06], P < .001). The 6-month mortality and hospitalization risks in patients with dementia and COVID-19 were 20.99% and 59.26%, respectively. DISCUSSION: These findings highlight the need to protect patients with dementia as part of the strategy to control the COVID-19 pandemic.

CoMNRank: An integrated approach to extract and prioritize human microbial metabolites from MEDLINE records.

MOTIVATION: Trillions of bacteria in human body (human microbiota) affect human health and diseases by controlling host functions through small molecule metabolites.An accurate and comprehensive catalog of the metabolic output from human microbiota is critical for our deep understanding of how microbial metabolism contributes to human health.The large number of published biomedical research articles is a rich resource of microbiome studies.However, automatically extracting microbial metabolites from free-text documents and differentiating them from other human metabolites is a challenging task.Here we developed an integrated approach called Co-occurrence Metabolite Network Ranking (CoMNRank) by combining named entity extraction, network construction and topic sensitive network-based prioritization to extract and prioritize microbial metabolites from biomedical articles. METHODS: The text data included 28,851,232 MEDLINE records.CoMNRank consists of three steps: (1) extraction of human metabolites from MEDLINE records; (2) construction of a weighted co-occurrence metabolite network (CoMN); (3) prioritization and differentiation of microbial metabolites from other human metabolites. RESULTS: For the first step of CoMNRank, we extracted 11,846 human metabolites from MEDLINE articles, with a baseline performance of precision of 0.014, recall of 0.959 and F1 of 0.028.We then constructed a weighted CoMN of 6,996 nodes and 986,186 edges.CoMNRank effectively prioritized microbial metabolites: the precision of top ranked metabolites is 0.45, a 31-fold enrichment as compared to the overall precision of 0.014.Manual curation of top 100 metabolites showed a true precision of 0.67, among which 48% true positives are not captured by existing databases. CONCLUSION: Our study sets the foundation for future tasks of microbial entity and relationship extractions as well as data-driven studies of how microbial metabolism contributes to human health and diseases.

Data-driven multiple-level analysis of gut-microbiome-immune-joint interactions in rheumatoid arthritis.

BACKGROUND: Rheumatoid arthritis (RA) is the most common autoimmune disease and affects about 1% of the population. The cause of RA remains largely unknown and could result from a complex interaction between genes and environment factors. Recent studies suggested that gut microbiota and their collective metabolic outputs exert profound effects on the host immune system and are implicated in RA. However, which and how gut microbial metabolites interact with host genetics in contributing to RA pathogenesis remains unknown. In this study, we present a data-driven study to understand how gut microbial metabolites contribute to RA at the genetic, functional and phenotypic levels. RESULTS: We used publicly available disease genetics, chemical genetics, human metabolome, genetic signaling pathways, mouse genome-wide mutation phenotypes, and mouse phenotype ontology data. We identified RA-associated microbial metabolites and prioritized them based on their genetic, functional and phenotypic relevance to RA. We evaluated the prioritization methods using short-chain fatty acids (SCFAs), which were previously shown to be involved in RA etiology. We validate the algorithms by showing that SCFAs are highly associated with RA at genetic, functional and phenotypic levels: SCFAs ranked at top 3.52% based on shared genes with RA, top 5.69% based on shared genetic pathways, and top 16.94% based on shared phenotypes. Based on the genetic-level analysis, human gut microbial metabolites directly interact with many RA-associated genes (as many as 18.1% of all 166 RA genes). Based on the functional-level analysis, human gut microbial metabolites participate in many RA-associated genetic pathways (as many as 71.4% of 311 genetic pathways significantly enriched for RA), including immune system pathways. Based on the phenotypic-level analysis, gut microbial metabolites affect many RA-related phenotypes (as many as 51.3% of 978 phenotypes significantly enriched for RA), including many immune system phenotypes. CONCLUSIONS: Our study demonstrates strong gut-microbiome-immune-joint interactions in RA, which converged on both genetic, functional and phenotypic levels.

Evaluating class III antiarrhythmic agents as novel MYC targeting drugs in ovarian cancer.

OBJECTIVE: To evaluate the utility of amiodarone and its derivative dronedarone as novel drug repositioning candidates in EOC and to determine the potential pathways targeted by these drugs. METHODS: Drug-predict bioinformatics platform was used to assess the utility of amiodarone as a novel drug-repurposing candidate in EOC. EOC cells were treated with amiodarone and dronedarone. Cell death was assessed by Annexin V staining. Cell viability and cell survival were assessed by MTT and clonogenics assays respectively. c-MYC and mTOR/Akt axis were evaluated as potential targets. Effect on autophagy was determined by autophagy flux flow cytometry. RESULTS: "DrugPredict" bioinformatics platform ranked Class III antiarrhythmic drug amiodarone within the top 3.9% of potential EOC drug repositioning candidates which was comparable to carboplatin ranking in the top 3.7%. Amiodarone and dronedarone were the only Class III antiarrhythmic drugs that decreased the cellular survival of both cisplatin-sensitive and cisplatin-resistant primary EOC cells. Interestingly, both drugs induced degradation of c-MYC protein and decreased the expression of known transcriptional targets of c-MYC. Furthermore, stable overexpression of non-degradable c-MYC partially rescued the effects of amiodarone and dronedarone induced cell death. Dronedarone induced higher autophagy flux in EOC cells as compared to amiodarone with decreased phospho-AKT and phospho-4EBP1 protein expression, suggesting autophagy induction due to inhibition of AKT/mTOR axis with these drugs. Lastly, both drugs also inhibited the survival of EOC tumor-initiating cells (TICs). CONCLUSIONS: We provide the first evidence of class III antiarrhythmic agents as novel c-MYC targeting drugs and autophagy inducers in EOC. Since c-MYC is amplified in >40% ovarian tumors, our results provide the basis for repositioning amiodarone and dronedarone as novel c-MYC targeting drugs in EOC with potential extension to other cancers.

MetabolitePredict: A de novo human metabolomics prediction system and its applications in rheumatoid arthritis.

Human metabolomics has great potential in disease mechanism understanding, early diagnosis, and therapy. Existing metabolomics studies are often based on profiling patient biofluids and tissue samples and are difficult owing to the challenges of sample collection and data processing. Here, we report an alternative approach and developed a computation-based prediction system, MetabolitePredict, for disease metabolomics biomarker prediction. We applied MetabolitePredict to identify metabolite biomarkers and metabolite targeting therapies for rheumatoid arthritis (RA), a last-lasting complex disease with multiple genetic and environmental factors involved. MetabolitePredict is a de novo prediction system. It first constructs a disease-specific genetic profile using genes and pathways data associated with an input disease. It then constructs genetic profiles for a total of 259,170 chemicals/metabolites using known chemical genetics and human metabolomic data. MetabolitePredict prioritizes metabolites for a given disease based on the genetic profile similarities between disease and metabolites. We evaluated MetabolitePredict using 63 known RA-associated metabolites. MetabolitePredict found 24 of the 63 metabolites (recall: 0.38) and ranked them highly (mean ranking: top 4.13%, median ranking: top 1.10%, P-value: 5.08E-19). MetabolitePredict performed better than an existing metabolite prediction system, PROFANCY, in predicting RA-associated metabolites (PROFANCY: recall: 0.31, mean ranking: 20.91%, median ranking: 16.47%, P-value: 3.78E-7). Short-chain fatty acids (SCFAs), the abundant metabolites of gut microbiota in the fermentation of fiber, ranked highly (butyrate, 0.03%; acetate, 0.05%; propionate, 0.38%). Finally, we established MetabolitePredict's potential in novel metabolite targeting for disease treatment: MetabolitePredict ranked highly three known metabolite inhibitors for RA treatments (methotrexate:0.25%; leflunomide: 0.56%; sulfasalazine: 0.92%). MetabolitePredict is a generalizable disease metabolite prediction system. The only required input to the system is a disease name or a set of disease-associated genes. The web-based MetabolitePredict is available at:http://xulab. CASE: edu/MetabolitePredict.

A genomics-based systems approach towards drug repositioning for rheumatoid arthritis.

BACKGROUND: Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by inflammation and destruction of synovial joints. RA affects up to 1 % of the population worldwide. Currently, there are no drugs that can cure RA or achieve sustained remission. The unknown cause of the disease represents a significant challenge in the drug development. In this study, we address this challenge by proposing an alternative drug discovery approach that integrates and reasons over genetic interrelationships between RA and other genetic diseases as well as a large amount of higher-level drug treatment data. We first constructed a genetic disease network using disease genetics data from Genome-Wide Association Studies (GWAS). We developed a network-based ranking algorithm to prioritize diseases genetically-related to RA (RA-related diseases). We then developed a drug prioritization algorithm to reposition drugs from RA-related diseases to treat RA. RESULTS: Our algorithm found 74 of the 80 FDA-approved RA drugs and ranked them highly (recall: 0.925, median ranking: 8.93 %), demonstrating the validity of our strategy. When compared to a study that used GWAS data to directly connect RA-associated genes to drug targets ("direct genetics-based" approach), our algorithm ("indirect genetics-based") achieved a comparable overall performance, but complementary precision and recall in retrospective validation (precision: 0.22, recall: 0.36; F1: 0.27 vs. precision: 0.74, recall: 0.16; F1: 0.28). Our approach performed significantly better in novel predictions when evaluated using 165 not-yet-FDA-approved RA drugs (precision: 0.46, recall: 0.50; F1: 0.47 vs. precision: 0.40, recall: 0.006; F1: 0.01). CONCLUSIONS: In summary, although the fundamental pathophysiological mechanisms remain uncharacterized, our proposed computation-based drug discovery approach to analyzing genetic and treatment interrelationships among thousands of diseases and drugs can facilitate the discovery of innovative drugs for treating RA.

Comparing a knowledge-driven approach to a supervised machine learning approach in large-scale extraction of drug-side effect relationships from free-text biomedical literature.

BACKGROUND: Systems approaches to studying drug-side-effect (drug-SE) associations are emerging as an active research area for both drug target discovery and drug repositioning. However, a comprehensive drug-SE association knowledge base does not exist. In this study, we present a novel knowledge-driven (KD) approach to effectively extract a large number of drug-SE pairs from published biomedical literature. DATA AND METHODS: For the text corpus, we used 21,354,075 MEDLINE records (119,085,682 sentences). First, we used known drug-SE associations derived from FDA drug labels as prior knowledge to automatically find SE-related sentences and abstracts. We then extracted a total of 49,575 drug-SE pairs from MEDLINE sentences and 180,454 pairs from abstracts. RESULTS: On average, the KD approach has achieved a precision of 0.335, a recall of 0.509, and an F1 of 0.392, which is significantly better than a SVM-based machine learning approach (precision: 0.135, recall: 0.900, F1: 0.233) with a 73.0% increase in F1 score. Through integrative analysis, we demonstrate that the higher-level phenotypic drug-SE relationships reflects lower-level genetic, genomic, and chemical drug mechanisms. In addition, we show that the extracted drug-SE pairs can be directly used in drug repositioning. CONCLUSION: In summary, we automatically constructed a large-scale higher-level drug phenotype relationship knowledge, which can have great potential in computational drug discovery.

A genome-wide systems analysis reveals strong link between colorectal cancer and trimethylamine N-oxide (TMAO), a gut microbial metabolite of dietary meat and fat.

BACKGROUND: Dietary intakes of red meat and fat are established risk factors for both colorectal cancer (CRC) and cardiovascular disease (CVDs). Recent studies have shown a mechanistic link between TMAO, an intestinal microbial metabolite of red meat and fat, and risk of CVDs. Data linking TMAO directly to CRC is, however, lacking. Here, we present an unbiased data-driven network-based systems approach to uncover a potential genetic relationship between TMAO and CRC. MATERIALS AND METHODS: We constructed two different epigenetic interaction networks (EINs) using chemical-gene, disease-gene and protein-protein interaction data from multiple large-scale data resources. We developed a network-based ranking algorithm to ascertain TMAO-related diseases from EINs. We systematically analyzed disease categories among TMAO-related diseases at different ranking cutoffs. We then determined which genetic pathways were associated with both TMAO and CRC. RESULTS: We show that CVDs and their major risk factors were ranked highly among TMAO-related diseases, confirming the newly discovered mechanistic link between CVDs and TMAO, and thus validating our algorithms. CRC was ranked highly among TMAO-related disease retrieved from both EINs (top 0.02%, #1 out of 4,372 diseases retrieved based on Mendelian genetics and top 10.9% among 882 diseases based on genome-wide association genetics), providing strong supporting evidence for our hypothesis that TMAO is genetically related to CRC. We have also identified putative genetic pathways that may link TMAO to CRC, which warrants further investigation. Through systematic disease enrichment analysis, we also demonstrated that TMAO is related to metabolic syndromes and cancers in general. CONCLUSIONS: Our genome-wide analysis demonstrates that systems approaches to studying the epigenetic interactions among diet, microbiome metabolisms, and disease genetics hold promise for understanding disease pathogenesis. Our results show that TMAO is genetically associated with CRC. This study suggests that TMAO may be an important intermediate marker linking dietary meat and fat and gut microbiota metabolism to risk of CRC, underscoring opportunities for the development of new gut microbiome-dependent diagnostic tests and therapeutics for CRC.

Combining automatic table classification and relationship extraction in extracting anticancer drug-side effect pairs from full-text articles.

Anticancer drug-associated side effect knowledge often exists in multiple heterogeneous and complementary data sources. A comprehensive anticancer drug-side effect (drug-SE) relationship knowledge base is important for computation-based drug target discovery, drug toxicity predication and drug repositioning. In this study, we present a two-step approach by combining table classification and relationship extraction to extract drug-SE pairs from a large number of high-profile oncological full-text articles. The data consists of 31,255 tables downloaded from the Journal of Oncology (JCO). We first trained a statistical classifier to classify tables into SE-related and -unrelated categories. We then extracted drug-SE pairs from SE-related tables. We compared drug side effect knowledge extracted from JCO tables to that derived from FDA drug labels. Finally, we systematically analyzed relationships between anti-cancer drug-associated side effects and drug-associated gene targets, metabolism genes, and disease indications. The statistical table classifier is effective in classifying tables into SE-related and -unrelated (precision: 0.711; recall: 0.941; F1: 0.810). We extracted a total of 26,918 drug-SE pairs from SE-related tables with a precision of 0.605, a recall of 0.460, and a F1 of 0.520. Drug-SE pairs extracted from JCO tables is largely complementary to those derived from FDA drug labels; as many as 84.7% of the pairs extracted from JCO tables have not been included a side effect database constructed from FDA drug labels. Side effects associated with anticancer drugs positively correlate with drug target genes, drug metabolism genes, and disease indications.

PhenoPredict: A disease phenome-wide drug repositioning approach towards schizophrenia drug discovery.

Schizophrenia (SCZ) is a common complex disorder with poorly understood mechanisms and no effective drug treatments. Despite the high prevalence and vast unmet medical need represented by the disease, many drug companies have moved away from the development of drugs for SCZ. Therefore, alternative strategies are needed for the discovery of truly innovative drug treatments for SCZ. Here, we present a disease phenome-driven computational drug repositioning approach for SCZ. We developed a novel drug repositioning system, PhenoPredict, by inferring drug treatments for SCZ from diseases that are phenotypically related to SCZ. The key to PhenoPredict is the availability of a comprehensive drug treatment knowledge base that we recently constructed. PhenoPredict retrieved all 18 FDA-approved SCZ drugs and ranked them highly (recall=1.0, and average ranking of 8.49%). When compared to PREDICT, one of the most comprehensive drug repositioning systems currently available, in novel predictions, PhenoPredict represented clear improvements over PREDICT in Precision-Recall (PR) curves, with a significant 98.8% improvement in the area under curve (AUC) of the PR curves. In addition, we discovered many drug candidates with mechanisms of action fundamentally different from traditional antipsychotics, some of which had published literature evidence indicating their treatment benefits in SCZ patients. In summary, although the fundamental pathophysiological mechanisms of SCZ remain unknown, integrated systems approaches to studying phenotypic connections among diseases may facilitate the discovery of innovative SCZ drugs.

Large-scale automatic extraction of side effects associated with targeted anticancer drugs from full-text oncological articles.

Targeted anticancer drugs such as imatinib, trastuzumab and erlotinib dramatically improved treatment outcomes in cancer patients, however, these innovative agents are often associated with unexpected side effects. The pathophysiological mechanisms underlying these side effects are not well understood. The availability of a comprehensive knowledge base of side effects associated with targeted anticancer drugs has the potential to illuminate complex pathways underlying toxicities induced by these innovative drugs. While side effect association knowledge for targeted drugs exists in multiple heterogeneous data sources, published full-text oncological articles represent an important source of pivotal, investigational, and even failed trials in a variety of patient populations. In this study, we present an automatic process to extract targeted anticancer drug-associated side effects (drug-SE pairs) from a large number of high profile full-text oncological articles. We downloaded 13,855 full-text articles from the Journal of Oncology (JCO) published between 1983 and 2013. We developed text classification, relationship extraction, signaling filtering, and signal prioritization algorithms to extract drug-SE pairs from downloaded articles. We extracted a total of 26,264 drug-SE pairs with an average precision of 0.405, a recall of 0.899, and an F1 score of 0.465. We show that side effect knowledge from JCO articles is largely complementary to that from the US Food and Drug Administration (FDA) drug labels. Through integrative correlation analysis, we show that targeted drug-associated side effects positively correlate with their gene targets and disease indications. In conclusion, this unique database that we built from a large number of high-profile oncological articles could facilitate the development of computational models to understand toxic effects associated with targeted anticancer drugs.

Large-scale combining signals from both biomedical literature and the FDA Adverse Event Reporting System (FAERS) to improve post-marketing drug safety signal detection.

BACKGROUND: Independent data sources can be used to augment post-marketing drug safety signal detection. The vast amount of publicly available biomedical literature contains rich side effect information for drugs at all clinical stages. In this study, we present a large-scale signal boosting approach that combines over 4 million records in the US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) and over 21 million biomedical articles. RESULTS: The datasets are comprised of 4,285,097 records from FAERS and 21,354,075 MEDLINE articles. We first extracted all drug-side effect (SE) pairs from FAERS. Our study implemented a total of seven signal ranking algorithms. We then compared these different ranking algorithms before and after they were boosted with signals from MEDLINE sentences or abstracts. Finally, we manually curated all drug-cardiovascular (CV) pairs that appeared in both data sources and investigated whether our approach can detect many true signals that have not been included in FDA drug labels. We extracted a total of 2,787,797 drug-SE pairs from FAERS with a low initial precision of 0.025. The ranking algorithm combined signals from both FAERS and MEDLINE, significantly improving the precision from 0.025 to 0.371 for top-ranked pairs, representing a 13.8 fold elevation in precision. We showed by manual curation that drug-SE pairs that appeared in both data sources were highly enriched with true signals, many of which have not yet been included in FDA drug labels. CONCLUSIONS: We have developed an efficient and effective drug safety signal ranking and strengthening approach We demonstrate that large-scale combining information from FAERS and biomedical literature can significantly contribute to drug safety surveillance.

dRiskKB: a large-scale disease-disease risk relationship knowledge base constructed from biomedical text.

BACKGROUND: Discerning the genetic contributions to complex human diseases is a challenging mandate that demands new types of data and calls for new avenues for advancing the state-of-the-art in computational approaches to uncovering disease etiology. Systems approaches to studying observable phenotypic relationships among diseases are emerging as an active area of research for both novel disease gene discovery and drug repositioning. Currently, systematic study of disease relationships on a phenome-wide scale is limited due to the lack of large-scale machine understandable disease phenotype relationship knowledge bases. Our study innovates a semi-supervised iterative pattern learning approach that is used to build an precise, large-scale disease-disease risk relationship (D1 → D2) knowledge base (dRiskKB) from a vast corpus of free-text published biomedical literature. RESULTS: 21,354,075 MEDLINE records comprised the text corpus under study. First, we used one typical disease risk-specific syntactic pattern (i.e. "D1 due to D2") as a seed to automatically discover other patterns specifying similar semantic relationships among diseases. We then extracted D1 → D2 risk pairs from MEDLINE using the learned patterns. We manually evaluated the precisions of the learned patterns and extracted pairs. Finally, we analyzed the correlations between disease-disease risk pairs and their associated genes and drugs. The newly created dRiskKB consists of a total of 34,448 unique D1 → D2 pairs, representing the risk-specific semantic relationships among 12,981 diseases with each disease linked to its associated genes and drugs. The identified patterns are highly precise (average precision of 0.99) in specifying the risk-specific relationships among diseases. The precisions of extracted pairs are 0.919 for those that are exactly matched and 0.988 for those that are partially matched. By comparing the iterative pattern approach starting from different seeds, we demonstrated that our algorithm is robust in terms of seed choice. We show that diseases and their risk diseases as well as diseases with similar risk profiles tend to share both genes and drugs. CONCLUSIONS: This unique dRiskKB, when combined with existing phenotypic, genetic, and genomic datasets, can have profound implications in our deeper understanding of disease etiology and in drug repositioning.

Towards building a disease-phenotype knowledge base: extracting disease-manifestation relationship from literature.

MOTIVATION: Systems approaches to studying phenotypic relationships among diseases are emerging as an active area of research for both novel disease gene discovery and drug repurposing. Currently, systematic study of disease phenotypic relationships on a phenome-wide scale is limited because large-scale machine-understandable disease-phenotype relationship knowledge bases are often unavailable. Here, we present an automatic approach to extract disease-manifestation (D-M) pairs (one specific type of disease-phenotype relationship) from the wide body of published biomedical literature. DATA AND METHODS: Our method leverages external knowledge and limits the amount of human effort required. For the text corpus, we used 119 085 682 MEDLINE sentences (21 354 075 citations). First, we used D-M pairs from existing biomedical ontologies as prior knowledge to automatically discover D-M-specific syntactic patterns. We then extracted additional pairs from MEDLINE using the learned patterns. Finally, we analysed correlations between disease manifestations and disease-associated genes and drugs to demonstrate the potential of this newly created knowledge base in disease gene discovery and drug repurposing. RESULTS: In total, we extracted 121 359 unique D-M pairs with a high precision of 0.924. Among the extracted pairs, 120 419 (99.2%) have not been captured in existing structured knowledge sources. We have shown that disease manifestations correlate positively with both disease-associated genes and drug treatments. CONCLUSIONS: The main contribution of our study is the creation of a large-scale and accurate D-M phenotype relationship knowledge base. This unique knowledge base, when combined with existing phenotypic, genetic and proteomic datasets, can have profound implications in our deeper understanding of disease etiology and in rapid drug repurposing. AVAILABILITY: http://nlp.case.edu/public/data/DMPatternUMLS/

Automatic signal extraction, prioritizing and filtering approaches in detecting post-marketing cardiovascular events associated with targeted cancer drugs from the FDA Adverse Event Reporting System (FAERS).

OBJECTIVE: Targeted drugs dramatically improve the treatment outcomes in cancer patients; however, these innovative drugs are often associated with unexpectedly high cardiovascular toxicity. Currently, cardiovascular safety represents both a challenging issue for drug developers, regulators, researchers, and clinicians and a concern for patients. While FDA drug labels have captured many of these events, spontaneous reporting systems are a main source for post-marketing drug safety surveillance in 'real-world' (outside of clinical trials) cancer patients. In this study, we present approaches to extracting, prioritizing, filtering, and confirming cardiovascular events associated with targeted cancer drugs from the FDA Adverse Event Reporting System (FAERS). DATA AND METHODS: The dataset includes records of 4,285,097 patients from FAERS. We first extracted drug-cardiovascular event (drug-CV) pairs from FAERS through named entity recognition and mapping processes. We then compared six ranking algorithms in prioritizing true positive signals among extracted pairs using known drug-CV pairs derived from FDA drug labels. We also developed three filtering algorithms to further improve precision. Finally, we manually validated extracted drug-CV pairs using 21 million published MEDLINE records. RESULTS: We extracted a total of 11,173 drug-CV pairs from FAERS. We showed that ranking by frequency is significantly more effective than by the five standard signal detection methods (246% improvement in precision for top-ranked pairs). The filtering algorithm we developed further improved overall precision by 91.3%. By manual curation using literature evidence, we show that about 51.9% of the 617 drug-CV pairs that appeared in both FAERS and MEDLINE sentences are true positives. In addition, 80.6% of these positive pairs have not been captured by FDA drug labeling. CONCLUSIONS: The unique drug-CV association dataset that we created based on FAERS could facilitate our understanding and prediction of cardiotoxic events associated with targeted cancer drugs.

Automatic construction of a large-scale and accurate drug-side-effect association knowledge base from biomedical literature.

Systems approaches to studying drug-side-effect (drug-SE) associations are emerging as an active research area for drug target discovery, drug repositioning, and drug toxicity prediction. However, currently available drug-SE association databases are far from being complete. Herein, in an effort to increase the data completeness of current drug-SE relationship resources, we present an automatic learning approach to accurately extract drug-SE pairs from the vast amount of published biomedical literature, a rich knowledge source of side effect information for commercial, experimental, and even failed drugs. For the text corpus, we used 119,085,682 MEDLINE sentences and their parse trees. We used known drug-SE associations derived from US Food and Drug Administration (FDA) drug labels as prior knowledge to find relevant sentences and parse trees. We extracted syntactic patterns associated with drug-SE pairs from the resulting set of parse trees. We developed pattern-ranking algorithms to prioritize drug-SE-specific patterns. We then selected a set of patterns with both high precisions and recalls in order to extract drug-SE pairs from the entire MEDLINE. In total, we extracted 38,871 drug-SE pairs from MEDLINE using the learned patterns, the majority of which have not been captured in FDA drug labels to date. On average, our knowledge-driven pattern-learning approach in extracting drug-SE pairs from MEDLINE has achieved a precision of 0.833, a recall of 0.407, and an F1 of 0.545. We compared our approach to a support vector machine (SVM)-based machine learning and a co-occurrence statistics-based approach. We show that the pattern-learning approach is largely complementary to the SVM- and co-occurrence-based approaches with significantly higher precision and F1 but lower recall. We demonstrated by correlation analysis that the extracted drug side effects correlate positively with both drug targets, metabolism, and indications.