No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Biomedical Knowledge Graph Embeddings with Negative Statements
Abstract
A knowledge graph is a powerful representation of real-world entities and their relations. The vast majority of these relations are defined as positive statements, but the importance of negative statements is increasingly recognized, especially under an Open World Assumption. Explicitly considering negative statements has been shown to improve performance on tasks such as entity summarization and question answering or domain-specific tasks such as protein function prediction. However, no attention has been given to the exploration of negative statements by knowledge graph embedding approaches despite the potential of negative statements to produce more accurate representations of entities in a knowledge graph.
We propose a novel approach, TrueWalks, to incorporate negative statements into the knowledge graph representation learning process. In particular, we present a novel walk-generation method that is able to not only differentiate between positive and negative statements but also take into account the semantic implications of negation in ontology-rich knowledge graphs. This is of particular importance for applications in the biomedical domain, where the inadequacy of embedding approaches regarding negative statements at the ontology level has been identified as a crucial limitation.
We evaluate TrueWalks in ontology-rich biomedical knowledge graphs in two different predictive tasks based on KG embeddings: protein-protein interaction prediction and gene-disease association prediction. We conduct an extensive analysis over established benchmarks and demonstrate that our method is able to improve the performance of knowledge graph embeddings on all tasks.
Deciphering the relationship between human proteins (genes) and phenotypes is one of the fundamental tasks in phenomics research. The Human Phenotype Ontology (HPO) builds upon a standardized logical vocabulary to describe the abnormal phenotypes encountered in human diseases and paves the way towards the computational analysis of their genetic causes. To date, many computational methods have been proposed to predict the HPO annotations of proteins. In this paper, we conduct a comprehensive review of the existing approaches to predicting HPO annotations of novel proteins, identifying missing HPO annotations, and prioritizing candidate proteins with respect to a certain HPO term. For each topic, we first give the formalized description of the problem, and then systematically revisit the published literatures highlighting their advantages and disadvantages, followed by the discussion on the challenges and promising future directions. In addition, we point out several potential topics to be worthy of exploration including the selection of negative HPO annotations and detecting HPO misannotations. We believe that this review will provide insight to the researchers in the field of computational phenotype analyses in terms of comprehending and developing novel prediction algorithms.
In this article, we provide a comprehensive introduction to knowledge graphs, which have recently garnered significant attention from both industry and academia in scenarios that require exploiting diverse, dynamic, large-scale collections of data. After some opening remarks, we motivate and contrast various graph-based data models, as well as languages used to query and validate knowledge graphs. We explain how knowledge can be represented and extracted using a combination of deductive and inductive techniques. We conclude with high-level future research directions for knowledge graphs.
Semantic embedding of knowledge graphs has been widely studied and used for prediction and statistical analysis tasks across various domains such as Natural Language Processing and the Semantic Web. However, less attention has been paid to developing robust methods for embedding OWL (Web Ontology Language) ontologies, which contain richer semantic information than plain knowledge graphs, and have been widely adopted in domains such as bioinformatics. In this paper, we propose a random walk and word embedding based ontology embedding method named , which encodes the semantics of an OWL ontology by taking into account its graph structure, lexical information and logical constructors. Our empirical evaluation with three real world datasets suggests that benefits from these three different aspects of an ontology in class membership prediction and class subsumption prediction tasks. Furthermore, often significantly outperforms the state-of-the-art methods in our experiments.
Cellular life depends on a complex web of functional associations between biomolecules. Among these associations, protein–protein interactions are particularly important due to their versatility, specificity and adaptability. The STRING database aims to integrate all known and predicted associations between proteins, including both physical interactions as well as functional associations. To achieve this, STRING collects and scores evidence from a number of sources: (i) automated text mining of the scientific literature, (ii) databases of interaction experiments and annotated complexes/pathways, (iii) computational interaction predictions from co-expression and from conserved genomic context and (iv) systematic transfers of interaction evidence from one organism to another. STRING aims for wide coverage; the upcoming version 11.5 of the resource will contain more than 14 000 organisms. In this update paper, we describe changes to the text-mining system, a new scoring-mode for physical interactions, as well as extensive user interface features for customizing, extending and sharing protein networks. In addition, we describe how to query STRING with genome-wide, experimental data, including the automated detection of enriched functionalities and potential biases in the user's query data. The STRING resource is available online, at https://string-db.org/.
The Gene Ontology Consortium (GOC) provides the most comprehensive resource currently available for computable knowledge regarding the functions of genes and gene products. Here, we report the advances of the consortium over the past two years. The new GO-CAM annotation framework was notably improved, and we formalized the model with a computational schema to check and validate the rapidly increasing repository of 2838 GO-CAMs. In addition, we describe the impacts of several collaborations to refine GO and report a 10% increase in the number of GO annotations, a 25% increase in annotated gene products, and over 9,400 new scientific articles annotated. As the project matures, we continue our efforts to review older annotations in light of newer findings, and, to maintain consistency with other on-tologies. As a result, 20 000 annotations derived from experimental data were reviewed, corresponding to 2.5% of experimental GO annotations. The website (http://geneontology.org) was redesigned for quick access to documentation, downloads and tools. To maintain an accurate resource and support trace-ability and reproducibility, we have made available a historical archive covering the past 15 years of GO data with a consistent format and file structure for both the ontology and annotations.
The Human Phenotype Ontology (HPO, https://hpo.jax.org) was launched in 2008 to provide a comprehensive logical standard to describe and computationally analyze phenotypic abnormalities found in human disease. The HPO is now a worldwide standard for phenotype exchange. The HPO has grown steadily since its inception due to considerable contributions from clinical experts and researchers from a diverse range of disciplines. Here, we present recent major extensions of the HPO for neurology, nephrology, immunology, pulmonology, newborn screening, and other areas. For example, the seizure subontology now reflects the International League Against Epilepsy (ILAE) guidelines and these enhancements have already shown clinical validity. We present new efforts to harmonize computational definitions of phenotypic abnormalities across the HPO and multiple phenotype ontologies used for animal models of disease. These efforts will benefit software such as Exomiser by improving the accuracy and scope of cross-species phenotype matching. The computational modeling strategy used by the HPO to define disease entities and phenotypic features and distinguish between them is explained in detail.We also report on recent efforts to translate the HPO into indigenous languages. Finally, we summarize recent advances in the use of HPO in electronic health record systems.
Ontologies have long been employed in the life sciences to formally represent and reason over domain knowledge and they are employed in almost every major biological database. Recently, ontologies are increasingly being used to provide background knowledge in similarity-based analysis and machine learning models. The methods employed to combine ontologies and machine learning are still novel and actively being developed. We provide an overview over the methods that use ontologies to compute similarity and incorporate them in machine learning methods; in particular, we outline how semantic similarity measures and ontology embeddings can exploit the background knowledge in ontologies and how ontologies can provide constraints that improve machine learning models. The methods and experiments we describe are available as a set of executable notebooks, and we also provide a set of slides and additional resources at https://github.com/bio-ontology-research-group/machine-learning-with-ontologies.
Motivation:
With the ever-increasing number and diversity of sequenced species, the challenge to characterize genes with functional information is even more important. In most species, this characterization almost entirely relies on automated electronic methods. As such, it is critical to benchmark the various methods. The Critical Assessment of protein Function Annotation algorithms (CAFA) series of community experiments provide the most comprehensive benchmark, with a time-delayed analysis leveraging newly curated experimentally supported annotations. However, the definition of a false positive in CAFA has not fully accounted for the open world assumption (OWA), leading to a systematic underestimation of precision. The main reason for this limitation is the relative paucity of negative experimental annotations.
Results:
This article introduces a new, OWA-compliant, benchmark based on a balanced test set of positive and negative annotations. The negative annotations are derived from expert-curated annotations of protein families on phylogenetic trees. This approach results in a large increase in the average information content of negative annotations. The benchmark has been tested using the naïve and BLAST baseline methods, as well as two orthology-based methods. This new benchmark could complement existing ones in future CAFA experiments.
Availability and implementation:
All data, as well as code used for analysis, is available from https://lab.dessimoz.org/20_not.
Supplementary information:
Supplementary data are available at Bioinformatics online.
Background:
In recent years, biomedical ontologies have become important for describing existing biological knowledge in the form of knowledge graphs. Data mining approaches that work with knowledge graphs have been proposed, but they are based on vector representations that do not capture the full underlying semantics. An alternative is to use machine learning approaches that explore semantic similarity. However, since ontologies can model multiple perspectives, semantic similarity computations for a given learning task need to be fine-tuned to account for this. Obtaining the best combination of semantic similarity aspects for each learning task is not trivial and typically depends on expert knowledge.
Results:
We have developed a novel approach, evoKGsim, that applies Genetic Programming over a set of semantic similarity features, each based on a semantic aspect of the data, to obtain the best combination for a given supervised learning task. The approach was evaluated on several benchmark datasets for protein-protein interaction prediction using the Gene Ontology as the knowledge graph to support semantic similarity, and it outperformed competing strategies, including manually selected combinations of semantic aspects emulating expert knowledge. evoKGsim was also able to learn species-agnostic models with different combinations of species for training and testing, effectively addressing the limitations of predicting protein-protein interactions for species with fewer known interactions.
Conclusions:
evoKGsim can overcome one of the limitations in knowledge graph-based semantic similarity applications: the need to expertly select which aspects should be taken into account for a given application. Applying this methodology to protein-protein interaction prediction proved successful, paving the way to broader applications.
One of the most pressing challenges in genomic medicine is to understand the role played by genetic variation in health and disease. Thanks to the exploration of genomic variants at large scale, hundreds of thousands of disease-associated loci have been uncovered. However, the identification of variants of clinical relevance is a significant challenge that requires comprehensive interrogation of previous knowledge and linkage to new experimental results. To assist in this complex task, we created DisGeNET (http://www.disgenet.org/), a knowledge management platform integrating and standardizing data about disease associated genes and variants from multiple sources, including the scientific literature. DisGeNET covers the full spectrum of human diseases as well as normal and abnormal traits. The current release covers more than 24 000 diseases and traits, 17 000 genes and 117 000 genomic variants. The latest developments of DisGeNET include new sources of data, novel data attributes and prioritization metrics, a redesigned web interface and recently launched APIs. Thanks to the data standardization, the combination of expert curated information with data automatically mined from the scientific literature, and a suite of tools for accessing its publicly available data, DisGeNET is an interoperable resource supporting a variety of applications in genomic medicine and drug R&D.
An embedding is a function that maps entities from one algebraic structure into another while preserving certain characteristics. Embeddings are being used successfully for mapping relational data or text into vector spaces where they can be used for machine learning, similarity search, or similar tasks. We address the problem of finding vector space embeddings for theories in the Description Logic ??⁺⁺ that are also models of the TBox. To find such embeddings, we define an optimization problem that characterizes the model-theoretic semantics of the operators in ??⁺⁺ within ℝⁿ, thereby solving the problem of finding an interpretation function for an ??⁺⁺ theory given a particular domain Δ. Our approach is mainly relevant to large ??⁺⁺ theories and knowledge bases such as the ontologies and knowledge graphs used in the life sciences. We demonstrate that our method can be used for improved prediction of protein--protein interactions when compared to semantic similarity measures or knowledge graph embeddings.
Background
Improving efficiency of disease diagnosis based on phenotype ontology is a critical yet challenging research area. Recently, Human Phenotype Ontology (HPO)-based semantic similarity has been affectively and widely used to identify causative genes and diseases. However, current phenotype similarity measurements just consider the annotations and hierarchy structure of HPO, neglecting the definition description of phenotype terms.
Results
In this paper, we propose a novel phenotype similarity measurement, termed as DisPheno, which adequately incorporates the definition of phenotype terms in addition to HPO structure and annotations to measure the similarity between phenotype terms. DisPheno also integrates phenotype term associations into phenotype-set similarity measurement using gene and disease annotations of phenotype terms.
Conclusions
Compared with five existing state-of-the-art methods, DisPheno shows great performance in HPO-based phenotype semantic similarity measurement and improves the efficiency of disease identification, especially on noisy patients dataset.
Complex diseases are known to be associated with disease genes. Uncovering disease-gene associations is critical for diagnosis, treatment, and prevention of diseases. Computational algorithms which effectively predict candidate disease-gene associations prior to experimental proof can greatly reduce the associated cost and time. Most existing methods are disease-specific which can only predict genes associated with a specific disease at a time. Similarities among diseases are not used during the prediction. Meanwhile, most methods predict new disease genes based on known associations, making them unable to predict disease genes for diseases without known associated genes.In this study, a manifold learning-based method is proposed for predicting disease-gene associations by assuming that the geodesic distance between any disease and its associated genes should be shorter than that of other non-associated disease-gene pairs. The model maps the diseases and genes into a lower dimensional manifold based on the known disease-gene associations, disease similarity and gene similarity to predict new associations in terms of the geodesic distance between disease-gene pairs. In the 3-fold cross-validation experiments, our method achieves scores of 0.882 and 0.854 in terms of the area under of the receiver operating characteristic (ROC) curve (AUC) for diseases with more than one known associated genes and diseases with only one known associated gene, respectively. Further de novo studies on Lung Cancer and Bladder Cancer also show that our model is capable of identifying new disease-gene associations.
Identifying disease genes from a vast amount of genetic data is one of the most challenging tasks in the post-genomic era. Also, complex diseases present highly heterogeneous genotype, which difficult biological marker identification. Machine learning methods are widely used to identify these markers, but their performance is highly dependent upon the size and quality of available data. In this study, we demonstrated that machine learning classifiers trained on gene functional similarities, using Gene Ontology (GO), can improve the identification of genes involved in complex diseases. For this purpose, we developed a supervised machine learning methodology to predict complex disease genes. The proposed pipeline was assessed using Autism Spectrum Disorder (ASD) candidate genes. A quantitative measure of gene functional similarities was obtained by employing different semantic similarity measures. To infer the hidden functional similarities between ASD genes, various types of machine learning classifiers were built on quantitative semantic similarity matrices of ASD and non-ASD genes. The classifiers trained and tested on ASD and non-ASD gene functional similarities outperformed previously reported ASD classifiers. For example, a Random Forest (RF) classifier achieved an AUC of 0. 80 for predicting new ASD genes, which was higher than the reported classifier (0.73). Additionally, this classifier was able to predict 73 novel ASD candidate genes that were enriched for core ASD phenotypes, such as autism and obsessive-compulsive behavior. In addition, predicted genes were also enriched for ASD co-occurring conditions, including Attention Deficit Hyperactivity Disorder (ADHD). We also developed a KNIME workflow with the proposed methodology which allows users to configure and execute it without requiring machine learning and programming skills. Machine learning is an effective and reliable technique to decipher ASD mechanism by identifying novel disease genes, but this study further demonstrated that their performance can be improved by incorporating a quantitative measure of gene functional similarities. Source code and the workflow of the proposed methodology are available at https://github.com/Muh-Asif/ASD-genes-prediction.
Motivation:
Most gene prioritization methods model each disease or phenotype individually, but this fails to capture patterns common to several diseases or phenotypes. To overcome this limitation, we formulate the gene prioritization task as the factorization of a sparsely filled gene-phenotype matrix, where the objective is to predict the unknown matrix entries. To deliver more accurate gene-phenotype matrix completion, we extend classical Bayesian matrix factorization to work with multiple side information sources. The availability of side information allows us to make non-trivial predictions for genes for which no previous disease association is known.
Results:
Our gene prioritization method can innovatively not only integrate data sources describing genes, but also data sources describing Human Phenotype Ontology terms. Experimental results on our benchmarks show that our proposed model can effectively improve accuracy over the well-established gene prioritization method, Endeavour. In particular, our proposed method offers promising results on diseases of the nervous system; diseases of the eye and adnexa; endocrine, nutritional and metabolic diseases; and congenital malformations, deformations and chromosomal abnormalities, when compared to Endeavour.
Availability and implementation:
The Bayesian data fusion method is implemented as a Python/C++ package: https://github.com/jaak-s/macau. It is also available as a Julia package: https://github.com/jaak-s/BayesianDataFusion.jl. All data and benchmarks generated or analyzed during this study can be downloaded at https://owncloud.esat.kuleuven.be/index.php/s/UGb89WfkZwMYoTn.
Supplementary information:
Supplementary data are available at Bioinformatics online.
We study the problem of representation learning in heterogeneous networks. Its unique challenges come from the existence of multiple types of nodes and links, which limit the feasibility of the conventional network embedding techniques. We develop two scalable representation learning models, namely metapath2vec and metapath2vec++. The metapath2vec model formalizes meta-path-based random walks to construct the heterogeneous neighborhood of a node and then leverages a heterogeneous skip-gram model to perform node embeddings. The metapath2vec++ model further enables the simultaneous modeling of structural and semantic correlations in heterogeneous networks. Extensive experiments show that metapath2vec and metapath2vec++ are able to not only outperform state-of-the-art embedding models in various heterogeneous network mining tasks, such as node classification, clustering, and similarity search, but also discern the structural and semantic correlations between diverse network objects.
The Gene Ontology (GO) is a formidable resource, but there are several considerations about it that are essential to understand the data and interpret it correctly. The GO is sufficiently simple that it can be used without deep understanding of its structure or how it is developed, which is both a strength and a weakness. In this chapter, we discuss some common misinterpretations of the ontology and the annotations. A better understanding of the pitfalls and the biases in the GO should help users make the most of this very rich resource. We also review some of the misconceptions and misleading assumptions commonly made about GO, including the effect of data incompleteness, the importance of annotation qualifiers, and the transitivity or lack thereof associated with different ontology relations. We also discuss several biases that can confound aggregate analyses such as gene enrichment analyses. For each of these pitfalls and biases, we suggest remedies and best practices.
Motivation:
Biological data and knowledge bases increasingly rely on Semantic Web technologies and the use of knowledge graphs for data integration, retrieval and federated queries. In the past years, feature learning methods that are applicable to graph-structured data are becoming available, but have not yet widely been applied and evaluated on structured biological knowledge. Results: We develop a novel method for feature learning on biological knowledge graphs. Our method combines symbolic methods, in particular knowledge representation using symbolic logic and automated reasoning, with neural networks to generate embeddings of nodes that encode for related information within knowledge graphs. Through the use of symbolic logic, these embeddings contain both explicit and implicit information. We apply these embeddings to the prediction of edges in the knowledge graph representing problems of function prediction, finding candidate genes of diseases, protein-protein interactions, or drug target relations, and demonstrate performance that matches and sometimes outperforms traditional approaches based on manually crafted features. Our method can be applied to any biological knowledge graph, and will thereby open up the increasing amount of Semantic Web based knowledge bases in biology to use in machine learning and data analytics.
Availability and implementation:
https://github.com/bio-ontology-research-group/walking-rdf-and-owl.
Contact:
robert.hoehndorf@kaust.edu.sa.
Supplementary information:
Supplementary data are available at Bioinformatics online.
In this work, we present a novel counter-fitting method which injects antonymy and synonymy constraints into vector space representations in order to improve the vectors' capability for judging semantic similarity. Applying this method to publicly available pre-trained word vectors leads to a new state of the art performance on the SimLex-999 dataset. We also show how the method can be used to tailor the word vector space for the downstream task of dialogue state tracking, resulting in robust improvements across different dialogue domains.
We present two simple modifications to the models in the popular Word2Vec tool, in order to generate embeddings more suited to tasks involving syntax. The main issue with the original models is the fact that they are insensitive to word order. While order independence is useful for inducing semantic representations, this leads to suboptimal results when they are used to solve syntax-based problems. We show improvements in part-of-speech tagging and dependency parsing using our proposed models.
Numerous new disease-gene associations have been identified by whole-exome sequencing studies in the last few years. However, many cases remain unsolved due to the sheer number of candidate variants remaining after common filtering strategies such as removing low quality and common variants and those deemed unlikely to be pathogenic. The observation that each of our genomes contains about 100 genuine loss-of-function variants makes identification of the causative mutation problematic when using these strategies alone. We propose using the wealth of genotype to phenotype data that already exists from model organism studies to assess the potential impact of these exome variants. Here, we introduce PHenotypic Interpretation of Variants in Exomes (PHIVE), an algorithm that integrates the calculation of phenotype similarity between human diseases and genetically modified mouse models with evaluation of the variants according to allele frequency, pathogenicity, and mode of inheritance approaches in our Exomiser tool. Large-scale validation of PHIVE analysis using 100,000 exomes containing known mutations demonstrated a substantial improvement (up to 54.1-fold) over purely variant-based (frequency and pathogenicity) methods with the correct gene recalled as the top hit in up to 83% of samples, corresponding to an area under the ROC curve of >95%. We conclude that incorporation of phenotype data can play a vital role in translational bioinformatics and propose that exome sequencing projects should systematically capture clinical phenotypes to take advantage of the strategy presented here.
We consider learning representations of entities and relations in KBs using
the neural-embedding approach. We show that most existing models, including NTN
and TransE, can be generalized under a unified learning framework, where
entities are low-dimensional vectors learned from a neural network and
relations are bilinear and/or linear mapping functions. Under this framework,
we compare a variety of embedding models on the link prediction task. We show
that a simple bilinear formulation achieves new state-of-the-art results for
the task (achieving a top-10 accuracy of 73.2% vs. 54.7% by TransE when
evaluated on Freebase). Furthermore, we introduce a novel approach that
utilizes the learned relation embeddings to mine logical rules from the KB. We
demonstrate that embeddings trained from the bilinear objective can effectively
capture relation composition via matrix multiplication. We also show that our
embedding-based approach can extract rules that involve relation transitivity
more effectively than a state-of-the-art rule mining approach that is tailored
for large-scale KBs.
We propose two novel model architectures for computing continuous vector
representations of words from very large data sets. The quality of these
representations is measured in a word similarity task, and the results are
compared to the previously best performing techniques based on different types
of neural networks. We observe large improvements in accuracy at much lower
computational cost, i.e. it takes less than a day to learn high quality word
vectors from a 1.6 billion words data set. Furthermore, we show that these
vectors provide state-of-the-art performance on our test set for measuring
syntactic and semantic word similarities.
Ontology management and maintenance are considered cor- nerstone issues in current Semantic Web applications in which semantic integration and ontological reasoning play a fundamental role. The ability to deal with inconsistency and to accommodate change is of utmost importance in real- world applications of ontological reasoning and management, wherein the need for expressing negated assertions also arises naturally. For this purpose, precise, formal definitions of the the different types of inconsistency and negation in ontologies are required. Unfortunately, ontology languages based on Description Logics (DLs) do not provide enough expressive power to represent axiom negations. Furthermore, there is no single, well-accepted notion of inconsistency and negation in the Semantic Web community, due to the lack of a common and solid foundational framework. In this paper, we propose a general framework accounting for inconsistency, negation and change in ontologies. Different levels of negation and inconsistency in DL-based ontologies are distinguished. We demonstrate how this framework can provide a foundation for reasoning with and management of dynamic ontologies.
A fundamental challenge in human health is the identification of disease-causing genes. Recently, several studies have tackled this challenge via a network-based approach, motivated by the observation that genes causing the same or similar diseases tend to lie close to one another in a network of protein-protein or functional interactions. However, most of these approaches use only local network information in the inference process and are restricted to inferring single gene associations. Here, we provide a global, network-based method for prioritizing disease genes and inferring protein complex associations, which we call PRINCE. The method is based on formulating constraints on the prioritization function that relate to its smoothness over the network and usage of prior information. We exploit this function to predict not only genes but also protein complex associations with a disease of interest. We test our method on gene-disease association data, evaluating both the prioritization achieved and the protein complexes inferred. We show that our method outperforms extant approaches in both tasks. Using data on 1,369 diseases from the OMIM knowledgebase, our method is able (in a cross validation setting) to rank the true causal gene first for 34% of the diseases, and infer 139 disease-related complexes that are highly coherent in terms of the function, expression and conservation of their member proteins. Importantly, we apply our method to study three multi-factorial diseases for which some causal genes have been found already: prostate cancer, alzheimer and type 2 diabetes mellitus. PRINCE's predictions for these diseases highly match the known literature, suggesting several novel causal genes and protein complexes for further investigation.
In recent years, ontologies have become a mainstream topic in biomedical research. When biological entities are described using a common schema, such as an ontology, they can be compared by means of their annotations. This type of comparison is called semantic similarity, since it assesses the degree of relatedness between two entities by the similarity in meaning of their annotations. The application of semantic similarity to biomedical ontologies is recent; nevertheless, several studies have been published in the last few years describing and evaluating diverse approaches. Semantic similarity has become a valuable tool for validating the results drawn from biomedical studies such as gene clustering, gene expression data analysis, prediction and validation of molecular interactions, and disease gene prioritization.
We review semantic similarity measures applied to biomedical ontologies and propose their classification according to the strategies they employ: node-based versus edge-based and pairwise versus groupwise. We also present comparative assessment studies and discuss the implications of their results. We survey the existing implementations of semantic similarity measures, and we describe examples of applications to biomedical research. This will clarify how biomedical researchers can benefit from semantic similarity measures and help them choose the approach most suitable for their studies.
Biomedical ontologies are evolving toward increased coverage, formality, and integration, and their use for annotation is increasingly becoming a focus of both effort by biomedical experts and application of automated annotation procedures to create corpora of higher quality and completeness than are currently available. Given that semantic similarity measures are directly dependent on these evolutions, we can expect to see them gaining more relevance and even becoming as essential as sequence similarity is today in biomedical research.
Recently, increasing efforts are put into learning continual representations for symbolic knowledge bases (KBs). However, these approaches either only embed the data-level knowledge (ABox) or suffer from inherent limitations when dealing with concept-level knowledge (TBox), i.e., they cannot faithfully model the logical structure present in the KBs. We present BoxEL, a geometric KB embedding approach that allows for better capturing the logical structure (i.e., ABox and TBox axioms) in the description logic . BoxEL models concepts in a KB as axis-parallel boxes that are suitable for modeling concept intersection, entities as points inside boxes, and relations between concepts/entities as affine transformations. We show theoretical guarantees (soundness) of BoxEL for preserving logical structure. Namely, the learned model of BoxEL embedding with loss 0 is a (logical) model of the KB. Experimental results on (plausible) subsumption reasonings and a real-world application–protein-protein prediction show that BoxEL outperforms traditional knowledge graph embedding methods as well as state-of-the-art embedding approaches.KeywordsOntologiesKnowledge graph embeddingsSemantic web
We deal with embedding a large scale knowledge graph composed of entities and relations into a continuous vector space. TransE is a promising method proposed recently, which is very efficient while achieving state-of-the-art predictive performance. We discuss some mapping properties of relations which should be considered in embedding, such as reflexive, one-to-many, many-to-one, and many-to-many. We note that TransE does not do well in dealing with these properties. Some complex models are capable of preserving these mapping properties but sacrifice efficiency in the process. To make a good trade-off between model capacity and efficiency, in this paper we propose TransH which models a relation as a hyperplane together with a translation operation on it. In this way, we can well preserve the above mapping properties of relations with almost the same model complexity of TransE. Additionally, as a practical knowledge graph is often far from completed, how to construct negative examples to reduce false negative labels in training is very important. Utilizing the one-to-many/many-to-one mapping property of a relation, we propose a simple trick to reduce the possibility of false negative labeling. We conduct extensive experiments on link prediction, triplet classification and fact extraction on benchmark datasets like WordNet and Freebase. Experiments show TransH delivers significant improvements over TransE on predictive accuracy with comparable capability to scale up.
Databases about general-world knowledge, so-called knowledge bases (KBs), are important in applications such as search and question answering. Traditionally, although KBs use open world assumption, popular KBs only store positive information, but withhold from taking any stance towards statements not contained in them. In this demo, we show that storing and presenting noteworthy negative statements would be important to overcome current limitations in various use cases. In particular, we introduce the Wiki neg ata portal, a platform to explore negative statements for Wikidata entities, by implementing a peer-based ranking method for inferring interesting negations in KBs. The demo is available at http://d5demos.mpi-inf.mpg.de/negation.
Knowledge bases (KBs) about notable entities and their properties are an important asset in applications such as search, question answering and dialog. All popular KBs capture virtually only positive statements, and abstain from taking any stance on statements not stored in the KB. This paper makes the case for explicitly stating salient statements that do not hold. Negative statements are useful to overcome limitations of question answering systems that are mainly geared for positive questions; they can also contribute to informative summaries of entities. Due to the abundance of such invalid statements, any effort to compile them needs to address ranking by saliency. We present a statistical inference method for compiling and ranking negative statements, based on expectations from positive statements of related entities in peer groups. Experimental results, with a variety of datasets, show that the method can effectively discover notable negative statements, and extrinsic studies underline their usefulness for entity summarization. Datasets and code are released as resources for further research.
Proteins interact with each other to play critical roles in many biological processes in cells. Although promising, laboratory experiments usually suffer from the disadvantages of being time-consuming and labor-intensive. The results obtained are often not robust and considerably uncertain. Due recently to advances in high-throughput technologies, a large amount of proteomics data has been collected and this presents a significant opportunity and also a challenge to develop computational models to predict protein–protein interactions (PPIs) based on these data. In this paper, we present a comprehensive survey of the recent efforts that have been made towards the development of effective computational models for PPI prediction. The survey introduces the algorithms that can be used to learn computational models for predicting PPIs, and it classifies these models into different categories. To understand their relative merits, the paper discusses different validation schemes and metrics to evaluate the prediction performance. Biological databases that are commonly used in different experiments for performance comparison are also described and their use in a series of extensive experiments to compare different prediction models are discussed. Finally, we present some open issues in PPI prediction for future work. We explain how the performance of PPI prediction can be improved if these issues are effectively tackled.
Complex biological systems are traditionally modelled as graphs of interconnected biological entities. These graphs, i.e. biological knowledge graphs, are then processed using graph exploratory approaches to perform different types of analytical and predictive tasks. Despite the high predictive accuracy of these approaches, they have limited scalability due to their dependency on time-consuming path exploratory procedures. In recent years, owing to the rapid advances of computational technologies, new approaches for modelling graphs and mining them with high accuracy and scalability have emerged. These approaches, i.e. knowledge graph embedding (KGE) models, operate by learning low-rank vector representations of graph nodes and edges that preserve the graph's inherent structure. These approaches were used to analyse knowledge graphs from different domains where they showed superior performance and accuracy compared to previous graph exploratory approaches. In this work, we study this class of models in the context of biological knowledge graphs and their different applications. We then show how KGE models can be a natural fit for representing complex biological knowledge modelled as graphs. We also discuss their predictive and analytical capabilities in different biology applications. In this regard, we present two example case studies that demonstrate the capabilities of KGE models: prediction of drug-target interactions and polypharmacy side effects. Finally, we analyse different practical considerations for KGEs, and we discuss possible opportunities and challenges related to adopting them for modelling biological systems.
The Human Phenotype Ontology (HPO) is a standardized set of phenotypic terms that are organized in a hierarchical fashion. It is a widely used resource for capturing human disease phenotypes for computational analysis to support differential diagnostics. The HPO is frequently used to create a set of terms that accurately describe the observed clinical abnormalities of an individual being evaluated for suspected rare genetic disease. This profile is compared with computational disease profiles in the HPO database with the aim of identifying genetic diseases with comparable phenotypic profiles. The computational analysis can be coupled with the analysis of whole‐exome or whole‐genome sequencing data through applications such as Exomiser. This article explains how to choose an optimal set of HPO terms for these cases and enter them with software, such as PhenoTips and PatientArchive, and demonstrates how to use Phenomizer and Exomiser to generate a computational differential diagnosis.
Motivation:
Ontologies are widely used in biology for data annotation, integration, and analysis. In addition to formally structured axioms, ontologies contain meta-data in the form of annotation axioms which provide valuable pieces of information that characterize ontology classes. Annotation axioms commonly used in ontologies include class labels, descriptions, or synonyms. Despite being a rich source of semantic information, the ontology meta-data are generally unexploited by ontology-based analysis methods such.
Results:
We propose a novel method, OPA2Vec, to generate vector representations of biological entities in ontologies by combining formal ontology axioms and annotation axioms from the ontology metadata. We apply a Word2Vec model that has been pre-trained on either a corpus or abstracts or full-text articles to produce feature vectors from our collected data. We validate our method in two different ways: first, we use the obtained vector representations of proteins in a similarity measure to predict protein-protein interaction on two different datasets. Second, we evaluate our method on predicting gene-disease associations based on phenotype similarity by generating vector representations of genes and diseases using a phenotype ontology, and applying the obtained vectors to predict gene-disease associations using mouse model phenotypes. We demonstrate that OPA2Vec significantly outperforms existing methods for predicting gene-disease associations. Using evidence from mouse models, we apply OPA2Vec to identify candidate genes for several thousand rare and orphan diseases. OPA2Vec can be used to produce vector representations of any biomedical entity given any type of biomedical ontology.
Availability:
https://github.com/bio-ontology-research-group/opa2vec.
Knowledge graph (KG) embedding is to embed components of a KG including entities and relations into continuous vector spaces, so as to simplify the manipulation while preserving the inherent structure of the KG. It can benefit a variety of downstream tasks such as KG completion and relation extraction, and hence has quickly gained massive attention. In this article, we provide a systematic review of existing techniques, including not only the state-of-the-arts but also those with latest trends. Particularly, we make the review based on the type of information used in the embedding task. Techniques that conduct embedding using only facts observed in the KG are first introduced. We describe the overall framework, specific model design, typical training procedures, as well as pros and cons of such techniques. After that, we discuss techniques that further incorporate additional information besides facts. We focus specifically on the use of entity types, relation paths, textual descriptions, and logical rules. Finally, we briefly introduce how KG embedding can be applied to and benefit a wide variety of downstream tasks such as KG completion, relation extraction, question answering, and so forth.
We propose two novel model architectures for computing continuous vector representations of words from very large data sets. The quality of these representations is measured in a word similarity task, and the results are compared to the previously best performing techniques based on different types of neural networks. We observe large improvements in accuracy at much lower computational cost, i.e. it takes less than a day to learn high quality word vectors from a 1.6 billion words data set. Furthermore, we show that these vectors provide state-of-the-art performance on our test set for measuring syntactic and semantic word similarities.
When investigating Mendelian disease using exome or genome sequencing, distinguishing disease-causing genetic variants from the multitude of candidate variants is a complex, multidimensional task. Many prioritization tools and online interpretation resources exist, and professional organizations have offered clinical guidelines for review and return of prioritization results. In this Review, we describe the strengths and weaknesses of widely used computational approaches, explain their roles in the diagnostic and discovery process and discuss how they can inform (and misinform) expert reviewers. We place variant prioritization in the wider context of gene prioritization, burden testing and genotype–phenotype association, and we discuss opportunities and challenges introduced by whole-genome sequencing.
Linked Open Data has been recognized as a valuable source for background information in data mining. However, most data mining tools require features in propositional form, i.e., a vector of nominal or numerical features associated with an instance, while Linked Open Data sources are graphs by nature. In this paper, we present RDF2Vec, an approach that uses language modeling approaches for unsupervised feature extraction from sequences of words, and adapts them to RDF graphs. We generate sequences by leveraging local information from graph sub-structures, harvested by Weisfeiler-Lehman Subtree RDF Graph Kernels and graph walks, and learn latent numerical representations of entities in RDF graphs. Our evaluation shows that such vector representations outperform existing techniques for the propositionalization of RDF graphs on a variety of different predictive machine learning tasks, and that feature vector representations of general knowledge graphs such as DBpedia and Wikidata can be easily reused for different tasks.
Prediction tasks over nodes and edges in networks require careful effort in engineering features used by learning algorithms. Recent research in the broader field of representation learning has led to significant progress in automating prediction by learning the features themselves. However, present feature learning approaches are not expressive enough to capture the diversity of connectivity patterns observed in networks. Here we propose node2vec, an algorithmic framework for learning continuous feature representations for nodes in networks. In node2vec, we learn a mapping of nodes to a low-dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes. We define a flexible notion of a node's network neighborhood and design a biased random walk procedure, which efficiently explores diverse neighborhoods. Our algorithm generalizes prior work which is based on rigid notions of network neighborhoods, and we argue that the added flexibility in exploring neighborhoods is the key to learning richer representations.
We demonstrate the efficacy of node2vec over existing state-of-the-art techniques on multi-label classification and link prediction in several real-world networks from diverse domains. Taken together, our work represents a new way for efficiently learning state-of-the-art task-independent representations in complex networks.
Motivation:
Predicting the biological functions of proteins is one of the key challenges in the post genomic era. Computational models have demonstrated the utility of applying machine learning methods to predict protein function. Most prediction methods explicitly require a set of negative examples- proteins that are known not carrying out a particular function. However, Gene Ontology (GO) almost always only provides the knowledge that proteins carry out a particular function, and functional annotations of proteins are incomplete. GO structurally organizes more than tens of thousands GO terms and a protein is annotated with several (or dozens) of these terms. For these reasons, the negative examples of a protein can greatly help distinguishing true positive examples of the protein from such a large candidate GO space.
Results:
In this paper, we present a novel approach (called NegGOA) to select negative examples. Specifically, NegGOA takes advantage of the ontology structure, available annotations and potentiality of additional annotations of a protein to choose negative examples of the protein. We compare NegGOA with other negative examples selection algorithms and find that NegGOA produces much fewer false negatives than them. We incorporate the selected negative examples into an efficient function prediction model to predict the functions of proteins in Yeast, Human, Mouse and Fly. NegGOA also demonstrates improved accuracy than these comparing algorithms across various evaluation metrics. In addition, NegGOA is less suffered from incomplete annotations of proteins than these comparing methods.
Availability:
The Matlab and R codes are available at https://sites.google.com/site/guoxian85/neggoa CONTACT: gxyu@swu.edu.cn SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Knowledge graph completion aims to perform link prediction between entities. In this paper, we consider the approach of knowledge graph embeddings. Recently, models such as TransE and TransH build entity and relation embeddings by regarding a relation as translation from head entity to tail entity. We note that these models simply put both entities and relations within the same semantic space. In fact, an entity may have multiple aspects and various relations may focus on different aspects of entities, which makes a common space insufficient for modeling. In this paper, we propose TransR to build entity and relation embeddings in separate entity space and relation spaces. Afterwards, we learn embeddings by first projecting entities from entity space to corresponding relation space and then building translations between projected entities. In experiments, we evaluate our models on three tasks including link prediction, triple classification and relational fact extraction. Experimental results show significant and consistent improvements compared to state-of-the-art baselines including TransE and TransH.
Identifying protein-protein interactions is important in molecular biology. Experimental methods to this issue have their limitations, and computational approaches have attracted more and more attentions from the biological community. The semantic similarity derived from the Gene Ontology (GO) annotation has been regarded as one of the most powerful indicators for protein interaction. However, conventional methods based on GO similarity fail to take advantage of the specificity of GO terms in the ontology graph. We proposed a GO-based method to predict protein-protein interaction by integrating different kinds of similarity measures derived from the intrinsic structure of GO graph. We extended five existing methods to derive the semantic similarity measures from the descending part of two GO terms in the GO graph, then adopted a feature integration strategy to combines both the ascending and the descending similarity scores derived from the three sub-ontologies to construct various kinds of features to characterize each protein pair. Support vector machines (SVM) were employed as discriminate classifiers, and five-fold cross validation experiments were conducted on both human and yeast protein-protein interaction datasets to evaluate the performance of different kinds of integrated features, the experimental results suggest the best performance of the feature that combines information from both the ascending and the descending parts of the three ontologies. Our method is appealing for effective prediction of protein-protein interaction.
We consider the problem of embedding entities and relationships of multi relational data in low-dimensional vector spaces. Our objective is to propose a canonical model which is easy to train, contains a reduced number of parameters and can scale up to very large databases. Hence, we propose TransE, a method which models relationships by interpreting them as translations operating on the low-dimensional embeddings of the entities. Despite its simplicity, this assumption proves to be powerful since extensive experiments show that TransE significantly outperforms state-of-the-art methods in link prediction on two knowledge bases. Besides, it can be successfully trained on a large scale data set with 1M entities, 25k relationships and more than 17M training samples.
We present DeepWalk, a novel approach for learning latent representations of vertices in a network. These latent representations encode social relations in a continuous vector space, which is easily exploited by statistical models. DeepWalk generalizes recent advancements in language modeling and unsupervised feature learning (or deep learning) from sequences of words to graphs.
DeepWalk uses local information obtained from truncated random walks to learn latent representations by treating walks as the equivalent of sentences. We demonstrate DeepWalk's latent representations on several multi-label network classification tasks for social networks such as BlogCatalog, Flickr, and YouTube. Our results show that DeepWalk outperforms challenging baselines which are allowed a global view of the network, especially in the presence of missing information. DeepWalk's representations can provide F1 scores up to 10% higher than competing methods when labeled data is sparse. In some experiments, DeepWalk's representations are able to outperform all baseline methods while using 60% less training data.
DeepWalk is also scalable. It is an online learning algorithm which builds useful incremental results, and is trivially parallelizable. These qualities make it suitable for a broad class of real world applications such as network classification, and anomaly detection.
In order to choose correctly the dimension of calibration model in chemistry, a new simple and effective method named Monte Carlo cross validation (MCCV) is introduced in the present work. Unlike leave-one-out procedure commonly used in chemometrics for cross validation (CV), the Monte Carlo cross validation developed in this paper is an asymptotically consistent method in determining the number of components in calibration model. It can avoid an unnecessary large model and therefore decreases the risk of over-fitting for the calibration model. The results obtained from simulation study showed that MCCV has an obviously larger probability than leave-one-out CV in choosing the correct number of components that the model should contain. The results from real data sets demonstrated that MCCV could successfully choose the appropriate model, but leave-one-out CV could not.
Since achieving W3C recommendation status in 2004, the Web Ontology Language (OWL) has been successfully applied to many problems in computer science. Practical experience with OWL has been quite positive in general; however, it has also revealed room for improvement in several areas. We systematically analyze the identified shortcomings of OWL, such as expressivity issues, problems with its syntaxes, and deficiencies in the definition of OWL species. Furthermore, we present an overview of OWL 2—an extension to and revision of OWL that is currently being developed within the W3C OWL Working Group. Many aspects of OWL have been thoroughly reengineered in OWL 2, thus producing a robust platform for future development of the language.
rdflib: a high level wrapper around the redland package for common rdf applications
- C Boettiger
Predicting gene-disease associations with knowledge graph embeddings over multiple ontologies
- S Nunes
- R T Sousa
- C Pesquita