Knowledge Graphs: Opportunities and Challenges

With the explosive growth of artificial intelligence (AI) and big data, it has become vitally important to organize and represent the enormous volume of knowledge appropriately. As graph data, knowledge graphs accumulate and convey knowledge of the real world. It has been well-recognized that knowledge graphs effectively represent complex information; hence, they rapidly gain the attention of academia and industry in recent years. Thus to develop a deeper understanding of knowledge graphs, this paper presents a systematic overview of this field. Specifically, we focus on the opportunities and challenges of knowledge graphs. We first review the opportunities of knowledge graphs in terms of two aspects: (1) AI systems built upon knowledge graphs; (2) potential application fields of knowledge graphs. Then, we thoroughly discuss severe technical challenges in this field, such as knowledge graph embeddings, knowledge acquisition, knowledge graph completion, knowledge fusion, and knowledge reasoning. We expect that this survey will shed new light on future research and the development of knowledge graphs.

Recent advances in machine learning applications in metabolic engineering

Metabolic engineering encompasses several widely-used strategies, which currently hold a high seat in the field of biotechnology when its potential is manifesting through a plethora of research and commercial products with a strong societal impact. The genomic revolution that occurred almost three decades ago has initiated the generation of large omics-datasets which has helped in gaining a better understanding of cellular behavior. The itinerary of metabolic engineering that has occurred based on these large datasets has allowed researchers to gain detailed insights and a reasonable understanding of the intricacies of biosystems. However, the existing trail-and-error approaches for metabolic engineering are laborious and time-intensive when it comes to the production of target compounds with high yields through genetic manipulations in host organisms. Machine learning (ML) coupled with the available metabolic engineering test instances and omics data brings a comprehensive and multidisciplinary approach that enables scientists to evaluate various parameters for effective strain design. This vast amount of biological data should be standardized through knowledge engineering to train different ML models for providing accurate predictions in gene circuits designing, modification of proteins, optimization of bioprocess parameters for scaling up, and screening of hyper-producing robust cell factories. This review briefs on the premise of ML, followed by mentioning various ML methods and algorithms alongside the numerous omics datasets available to train ML models for predicting metabolic outcomes with high-accuracy. The combinative interplay between the ML algorithms and biological datasets through knowledge engineering have guided the recent advancements in applications such as CRISPR/Cas systems, gene circuits, protein engineering, metabolic pathway reconstruction, and bioprocess engineering. Finally, this review addresses the probable challenges of applying ML in metabolic engineering which will guide the researchers toward novel techniques to overcome the limitations.

Machine learning performance validation and training using a 'perfect' expert system

A method is proposed for generating application domain agnostic data for training and evaluating machine learning systems. The proposed method randomly generates an expert system network based upon user specified parameters. This expert system serves as a generic model of an unspecified phenomena. The expert system is run to determine the ideal output from a set of random inputs. These inputs and ideal output are used for training and testing a machine learning system. This allows a machine learning technology to be developed and tested without requiring compatible test data to be collected or before data collection as a proof-of-concept validation of system operations. It also allows systems to be tested without data error noise or with known levels of noise and with other perturbations, to facilitate analysis. It may also facilitate testing system security, adversarial attacks and conducting other types of research into machine learning systems.

•

Provides an application domain agnostic way to test machine learning technologies and facilitates the generalization of results.

•

Allows technologies to be tested with data with different characteristics without having to locate datasets that have these characteristics.

•

Utilizes randomly generated network to represent non-specific phenomena which can be used for training and testing machine learning techniques.

Cross Knowledge-based Generative Zero-Shot Learning approach with Taxonomy Regularization

Although zero-shot learning (ZSL) has an inferential capability of recognizing new classes that have never been seen before, it always faces two fundamental challenges of the cross modality and cross-domain challenges. In order to alleviate these problems, we develop a generative network-based ZSL approach equipped with the proposed Cross Knowledge Learning (CKL) scheme and Taxonomy Regularization (TR). In our approach, the semantic features are taken as inputs, and the output is the synthesized visual features generated from the corresponding semantic features. CKL enables more relevant semantic features to be trained for semantic-to-visual feature embedding in ZSL, while Taxonomy Regularization (TR) significantly improves the intersections with unseen images with more generalized visual features generated from generative network. Extensive experiments on several benchmark datasets (i.e., AwA1, AwA2, CUB, NAB and aPY) show that our approach is superior to these state-of-the-art methods in terms of ZSL image classification and retrieval.

An open-source framework for neuroscience metadata management applied to digital reconstructions of neuronal morphology

Research advancements in neuroscience entail the production of a substantial amount of data requiring interpretation, analysis, and integration. The complexity and diversity of neuroscience data necessitate the development of specialized databases and associated standards and protocols. NeuroMorpho.Org is an online repository of over one hundred thousand digitally reconstructed neurons and glia shared by hundreds of laboratories worldwide. Every entry of this public resource is associated with essential metadata describing animal species, anatomical region, cell type, experimental condition, and additional information relevant to contextualize the morphological content. Until recently, the lack of a user-friendly, structured metadata annotation system relying on standardized terminologies constituted a major hindrance in this effort, limiting the data release pace. Over the past 2 years, we have transitioned the original spreadsheet-based metadata annotation system of NeuroMorpho.Org to a custom-developed, robust, web-based framework for extracting, structuring, and managing neuroscience information. Here we release the metadata portal publicly and explain its functionality to enable usage by data contributors. This framework facilitates metadata annotation, improves terminology management, and accelerates data sharing. Moreover, its open-source development provides the opportunity of adapting and extending the code base to other related research projects with similar requirements. This metadata portal is a beneficial web companion to NeuroMorpho.Org which saves time, reduces errors, and aims to minimize the barrier for direct knowledge sharing by domain experts. The underlying framework can be progressively augmented with the integration of increasingly autonomous machine intelligence components.

Architecture and usability of OntoKeeper, an ontology evaluation tool

Background

The existing community-wide bodies of biomedical ontologies are known to contain quality and content problems. Past research has revealed various errors related to their semantics and logical structure. Automated tools may help to ease the ontology construction, maintenance, assessment and quality assurance processes. However, there are relatively few tools that exist that can provide this support to knowledge engineers.

Method

We introduce OntoKeeper as a web-based tool that can automate quality scoring for ontology developers. We enlisted 5 experienced ontologists to test the tool and then administered the System Usability Scale to measure their assessment.

Results

In this paper, we present usability results from 5 ontologists revealing high system usability of OntoKeeper, and use-cases that demonstrate its capabilities in previous published biomedical ontology research.

Conclusion

To the best of our knowledge, OntoKeeper is the first of a few ontology evaluation tools that can help provide ontology evaluation functionality for knowledge engineers with good usability.

HemOnc: A new standard vocabulary for chemotherapy regimen representation in the OMOP common data model

Systematic application of observational data to the understanding of impacts of cancer treatments requires detailed information models allowing meaningful comparisons between treatment regimens. Unfortunately, details of systemic therapies are scarce in registries and data warehouses, primarily due to the complex nature of the protocols and a lack of standardization. Since 2011, we have been creating a curated and semi-structured website of chemotherapy regimens, HemOnc.org. In coordination with the Observational Health Data Sciences and Informatics (OHDSI) Oncology Subgroup, we have transformed a substantial subset of this content into the OMOP common data model, with bindings to multiple external vocabularies, e.g., RxNorm and the National Cancer Institute Thesaurus. Currently, there are >73,000 concepts and >177,000 relationships in the full vocabulary. Content related to the definition and composition of chemotherapy regimens has been released within the ATHENA tool (athena.ohdsi.org) for widespread utilization by the OHDSI membership. Here, we describe the rationale, data model, and initial contents of the HemOnc vocabulary along with several use cases for which it may be valuable.

PGxO and PGxLOD: a reconciliation of pharmacogenomic knowledge of various provenances, enabling further comparison

Background

Pharmacogenomics (PGx) studies how genomic variations impact variations in drug response phenotypes. Knowledge in pharmacogenomics is typically composed of units that have the form of ternary relationships gene variant – drug – adverse event. Such a relationship states that an adverse event may occur for patients having the specified gene variant and being exposed to the specified drug. State-of-the-art knowledge in PGx is mainly available in reference databases such as PharmGKB and reported in scientific biomedical literature. But, PGx knowledge can also be discovered from clinical data, such as Electronic Health Records (EHRs), and in this case, may either correspond to new knowledge or confirm state-of-the-art knowledge that lacks “clinical counterpart” or validation. For this reason, there is a need for automatic comparison of knowledge units from distinct sources.

Results

In this article, we propose an approach, based on Semantic Web technologies, to represent and compare PGx knowledge units. To this end, we developed PGxO, a simple ontology that represents PGx knowledge units and their components. Combined with PROV-O, an ontology developed by the W3C to represent provenance information, PGxO enables encoding and associating provenance information to PGx relationships. Additionally, we introduce a set of rules to reconcile PGx knowledge, i.e. to identify when two relationships, potentially expressed using different vocabularies and levels of granularity, refer to the same, or to different knowledge units. We evaluated our ontology and rules by populating PGxO with knowledge units extracted from PharmGKB (2701), the literature (65,720) and from discoveries reported in EHR analysis studies (only 10, manually extracted); and by testing their similarity. We called PGxLOD (PGx Linked Open Data) the resulting knowledge base that represents and reconciles knowledge units of those various origins.

Conclusions

The proposed ontology and reconciliation rules constitute a first step toward a more complete framework for knowledge comparison in PGx. In this direction, the experimental instantiation of PGxO, named PGxLOD, illustrates the ability and difficulties of reconciling various existing knowledge sources.

Electronic supplementary material

The online version of this article (10.1186/s12859-019-2693-9) contains supplementary material, which is available to authorized users.

Visualized Emotion Ontology: a model for representing visual cues of emotions

BACKGROUND

Healthcare services, particularly in patient-provider interaction, often involve highly emotional situations, and it is important for physicians to understand and respond to their patients' emotions to best ensure their well-being.

METHODS

In order to model the emotion domain, we have created the Visualized Emotion Ontology (VEO) to provide a semantic definition of 25 emotions based on established models, as well as visual representations of emotions utilizing shapes, lines, and colors.

RESULTS

As determined by ontology evaluation metrics, VEO exhibited better machine-readability (z=1.12), linguistic quality (z=0.61), and domain coverage (z=0.39) compared to a sample of cognitive ontologies. Additionally, a survey of 1082 participants through Amazon Mechanical Turk revealed that a significantly higher proportion of people agree than disagree with 17 out of our 25 emotion images, validating the majority of our visualizations.

CONCLUSION

From the development, evaluation, and serialization of the VEO, we have defined a set of 25 emotions using OWL that linked surveyed visualizations to each emotion. In the future, we plan to use the VEO in patient-facing software tools, such as embodied conversational agents, to enhance interactions between patients and providers in a clinical environment.

Semantic interoperability challenges to process large amount of data perspectives in forensic and legal medicine

This article is a position paper dealing with semantic interoperability challenges. It addresses the Variety and Veracity dimensions when integrating, sharing and reusing large amount of heterogeneous data for data analysis and decision making applications in the healthcare domain. Many issues are raised by the necessity to conform Big Data to interoperability standards. We discuss how semantics can contribute to the improvement of information sharing and address the problem of data mediation with domain ontologies. We then introduce the main steps for building domain ontologies as they could be implemented in the context of Forensic and Legal medicine. We conclude with a particular emphasis on the current limitations in standardisation and the importance of knowledge formalization.

Synthetic biology for manufacturing chemicals: constraints drive the use of non-conventional microbial platforms

Genetically modified microbes have had much industrial success producing protein-based products (such as antibodies and enzymes). However, engineering microbial workhorses for biomanufacturing of commodity compounds remains challenging. First, microbes cannot afford burdens with both overexpression of multiple enzymes and metabolite drainage for product synthesis. Second, synthetic circuits and introduced heterologous pathways are not yet as "robust and reliable" as native pathways due to hosts' innate regulations, especially under suboptimal fermentation conditions. Third, engineered enzymes may lack channeling capabilities for cascade-like transport of metabolites to overcome diffusion barriers or to avoid intermediate toxicity in the cytoplasmic environment. Fourth, moving engineered hosts from laboratory to industry is unreliable because genetic mutations and non-genetic cell-to-cell variations impair the large-scale fermentation outcomes. Therefore, synthetic biology strains often have unsatisfactory industrial performance (titer/yield/productivity). To overcome these problems, many different species are being explored for their metabolic strengths that can be leveraged to synthesize specific compounds. Here, we provide examples of non-conventional and genetically amenable species for industrial manufacturing, including the following: Corynebacterium glutamicum for its TCA cycle-derived biosynthesis, Yarrowia lipolytica for its biosynthesis of fatty acids and carotenoids, cyanobacteria for photosynthetic production from its sugar phosphate pathways, and Rhodococcus for its ability to biotransform recalcitrant feedstock. Finally, we discuss emerging technologies (e.g., genome-to-phenome mapping, single cell methods, and knowledge engineering) that may facilitate the development of novel cell factories.

Transitive closure of subsumption and causal relations in a large ontology of radiological diagnosis

The Radiology Gamuts Ontology (RGO)-an ontology of diseases, interventions, and imaging findings-was developed to aid in decision support, education, and translational research in diagnostic radiology. The ontology defines a subsumption (is_a) relation between more general and more specific terms, and a causal relation (may_cause) to express the relationship between disorders and their possible imaging manifestations. RGO incorporated 19,745 terms with their synonyms and abbreviations, 1768 subsumption relations, and 55,558 causal relations. Transitive closure was computed iteratively; it yielded 2154 relations over subsumption and 1,594,896 relations over causality. Five causal cycles were discovered, all with path length of no more than 5. The graph-theoretic metrics of in-degree and out-degree were explored; the most useful metric to prioritize modification of the ontology was found to be the product of the in-degree of transitive closure over subsumption and the out-degree of transitive closure over causality. Two general types of error were identified: (1) causal assertions that used overly general terms because they implicitly assumed an organ-specific context and (2) subsumption relations where a site-specific disorder was asserted to be a subclass of the general disorder. Transitive closure helped identify incorrect assertions, prioritized and guided ontology revision, and aided resources that applied the ontology's knowledge.

Integrating electronic health record information to support integrated care: practical application of ontologies to improve the accuracy of diabetes disease registers

BACKGROUND

Information in Electronic Health Records (EHRs) are being promoted for use in clinical decision support, patient registers, measurement and improvement of integration and quality of care, and translational research. To do this EHR-derived data product creators need to logically integrate patient data with information and knowledge from diverse sources and contexts.

OBJECTIVE

To examine the accuracy of an ontological multi-attribute approach to create a Type 2 Diabetes Mellitus (T2DM) register to support integrated care.

METHODS

Guided by Australian best practice guidelines, the T2DM diagnosis and management ontology was conceptualized, contextualized and validated by clinicians; it was then specified, formalized and implemented. The algorithm was standardized against the domain ontology in SNOMED CT-AU. Accuracy of the implementation was measured in 4 datasets of varying sizes (927-12,057 patients) and an integrated dataset (23,793 patients). Results were cross-checked with sensitivity and specificity calculated with 95% confidence intervals.

RESULTS

Incrementally integrating Reason for Visit (RFV), medication (Rx), and pathology in the algorithm identified nearly100% of T2DM cases. Incrementally integrating the four datasets improved accuracy; controlling for sample size, data incompleteness and duplicates. Manual validation confirmed the accuracy of the algorithm.

CONCLUSION

Integrating multiple data elements within an EHR using ontology-based case-finding algorithms can improve the accuracy of the diagnosis and compensate for suboptimal data quality, and hence creating a dataset that is more fit-for-purpose. This clinical and pragmatic application of ontologies to EHR data improves the integration of data and the potential for better use of data to improve the quality of care.

International data-sharing for radiotherapy research: an open-source based infrastructure for multicentric clinical data mining

Extensive, multifactorial data sharing is a crucial prerequisite for current and future (radiotherapy) research. However, the cost, time and effort to achieve this are often a roadblock. We present an open-source based data-sharing infrastructure between two radiotherapy departments, allowing seamless exchange of de-identified, automatically translated clinical and biomedical treatment data.