TopoICSim: a new semantic similarity measure based on gene ontology

Coherent pathway enrichment estimation by modeling inter-pathway dependencies using regularized regression

MOTIVATION

Gene set enrichment methods are a common tool to improve the interpretability of gene lists as obtained, for example, from differential gene expression analyses. They are based on computing whether dysregulated genes are located in certain biological pathways more often than expected by chance. Gene set enrichment tools rely on pre-existing pathway databases such as KEGG, Reactome, or the Gene Ontology. These databases are increasing in size and in the number of redundancies between pathways, which complicates the statistical enrichment computation.

RESULTS

We address this problem and develop a novel gene set enrichment method, called pareg, which is based on a regularized generalized linear model and directly incorporates dependencies between gene sets related to certain biological functions, for example, due to shared genes, in the enrichment computation. We show that pareg is more robust to noise than competing methods. Additionally, we demonstrate the ability of our method to recover known pathways as well as to suggest novel treatment targets in an exploratory analysis using breast cancer samples from TCGA.

AVAILABILITY AND IMPLEMENTATION

pareg is freely available as an R package on Bioconductor (https://bioconductor.org/packages/release/bioc/html/pareg.html) as well as on https://github.com/cbg-ethz/pareg. The GitHub repository also contains the Snakemake workflows needed to reproduce all results presented here.

HiG2Vec: hierarchical representations of Gene Ontology and genes in the Poincaré ball

MOTIVATION

Knowledge manipulation of Gene Ontology (GO) and Gene Ontology Annotation (GOA) can be done primarily by using vector representation of GO terms and genes. Previous studies have represented GO terms and genes or gene products in Euclidean space to measure their semantic similarity using an embedding method such as the Word2Vec-based method to represent entities as numeric vectors. However, this method has the limitation that embedding large graph-structured data in the Euclidean space cannot prevent a loss of information of latent hierarchies, thus precluding the semantics of GO and GOA from being captured optimally. On the other hand, hyperbolic spaces such as the Poincaré balls are more suitable for modeling hierarchies, as they have a geometric property in which the distance increases exponentially as it nears the boundary because of negative curvature.

RESULTS

In this article, we propose hierarchical representations of GO and genes (HiG2Vec) by applying Poincaré embedding specialized in the representation of hierarchy through a two-step procedure: GO embedding and gene embedding. Through experiments, we show that our model represents the hierarchical structure better than other approaches and predicts the interaction of genes or gene products similar to or better than previous studies. The results indicate that HiG2Vec is superior to other methods in capturing the GO and gene semantics and in data utilization as well. It can be robustly applied to manipulate various biological knowledge.

AVAILABILITYAND IMPLEMENTATION

https://github.com/JaesikKim/HiG2Vec.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Predicting metabolic pathway membership with deep neural networks by integrating sequential and ontology information

BACKGROUND

Inference of protein's membership in metabolic pathways has become an important task in functional annotation of protein. The membership information can provide valuable context to the basic functional annotation and also aid reconstruction of incomplete pathways. Previous works have shown success of inference by using various similarity measures of gene ontology.

RESULTS

In this work, we set out to explore integrating ontology and sequential information to further improve the accuracy. Specifically, we developed a neural network model with an architecture tailored to facilitate the integration of features from different sources. Furthermore, we built models that are able to perform predictions from pathway-centric or protein-centric perspectives. We tested the classifiers using 5-fold cross validation for all metabolic pathways reported in KEGG database.

CONCLUSIONS

The testing results demonstrate that by integrating ontology and sequential information with a tailored architecture our deep neural network method outperforms the existing methods significantly in the pathway-centric mode, and in the protein-centric mode, our method either outperforms or performs comparably with a suite of existing GO term based semantic similarity methods.

Mantis: flexible and consensus-driven genome annotation

Background

The rapid development of the (meta-)omics fields has produced an unprecedented amount of high-resolution and high-fidelity data. Through the use of these datasets we can infer the role of previously functionally unannotated proteins from single organisms and consortia. In this context, protein function annotation can be described as the identification of regions of interest (i.e., domains) in protein sequences and the assignment of biological functions. Despite the existence of numerous tools, challenges remain in terms of speed, flexibility, and reproducibility. In the big data era, it is also increasingly important to cease limiting our findings to a single reference, coalescing knowledge from different data sources, and thus overcoming some limitations in overly relying on computationally generated data from single sources.

Results

We implemented a protein annotation tool, Mantis, which uses database identifiers intersection and text mining to integrate knowledge from multiple reference data sources into a single consensus-driven output. Mantis is flexible, allowing for the customization of reference data and execution parameters, and is reproducible across different research goals and user environments. We implemented a depth-first search algorithm for domain-specific annotation, which significantly improved annotation performance compared to sequence-wide annotation. The parallelized implementation of Mantis results in short runtimes while also outputting high coverage and high-quality protein function annotations.

Conclusions

Mantis is a protein function annotation tool that produces high-quality consensus-driven protein annotations. It is easy to set up, customize, and use, scaling from single genomes to large metagenomes. Mantis is available under the MIT license at https://github.com/PedroMTQ/mantis.

GAPGOM-an R package for gene annotation prediction using GO Metrics

Objective

Properties of gene products can be described or annotated with Gene Ontology (GO) terms. But for many genes we have limited information about their products, for example with respect to function. This is particularly true for long non-coding RNAs (lncRNAs), where the function in most cases is unknown. However, it has been shown that annotation as described by GO terms to some extent can be predicted by enrichment analysis on properties of co-expressed genes.

Results

GAPGOM integrates two relevant algorithms, lncRNA2GOA and TopoICSim, into a user-friendly R package. Here lncRNA2GOA does annotation prediction by co-expression, whereas TopoICSim estimates similarity between GO graphs, which can be used for benchmarking of prediction performance, but also for comparison of GO graphs in general. The package provides an improved implementation of the original tools, with substantial improvements in performance and documentation, unified interfaces, and additional features.

Influence of the go-based semantic similarity measures in multi-objective gene clustering algorithm performance

Using a prior biological knowledge of relationships and genetic functions for gene similarity, from repository such as the Gene Ontology (GO), has shown good results in multi-objective gene clustering algorithms. In this scenario and to obtain useful clustering results, it would be helpful to know which measure of biological similarity between genes should be employed to yield meaningful clusters that have both similar expression patterns (co-expression) and biological homogeneity. In this paper, we studied the influence of the four most used GO-based semantic similarity measures in the performance of a multi-objective gene clustering algorithm. We used four publicly available datasets and carried out comparative studies based on performance metrics for the multi-objective optimization field and clustering performance indexes. In most of the cases, using Jiang-Conrath and Wang similarities stand in terms of multi-objective metrics. In clustering properties, Resnik similarity allows to achieve the best values of compactness and separation and therefore of co-expression of groups of genes. Meanwhile, in biological homogeneity, the Wang similarity reports greater number of significant GO terms. However, statistical, visual, and biological significance tests showed that none of the GO-based semantic similarity measures stand out above the rest in order to significantly improve the performance of the multi-objective gene clustering algorithm.

A Collection of Benchmark Data Sets for Knowledge Graph-based Similarity in the Biomedical Domain

The ability to compare entities within a knowledge graph is a cornerstone technique for several applications, ranging from the integration of heterogeneous data to machine learning. It is of particular importance in the biomedical domain, where semantic similarity can be applied to the prediction of protein-protein interactions, associations between diseases and genes, cellular localization of proteins, among others. In recent years, several knowledge graph-based semantic similarity measures have been developed, but building a gold standard data set to support their evaluation is non-trivial. We present a collection of 21 benchmark data sets that aim at circumventing the difficulties in building benchmarks for large biomedical knowledge graphs by exploiting proxies for biomedical entity similarity. These data sets include data from two successful biomedical ontologies, Gene Ontology and Human Phenotype Ontology, and explore proxy similarities calculated based on protein sequence similarity, protein family similarity, protein-protein interactions and phenotype-based gene similarity. Data sets have varying sizes and cover four different species at different levels of annotation completion. For each data set, we also provide semantic similarity computations with state-of-the-art representative measures. Database URL: https://github.com/liseda-lab/kgsim-benchmark.

Handling Big Data Scalability in Biological Domain Using Parallel and Distributed Processing: A Case of Three Biological Semantic Similarity Measures

In the field of biology, researchers need to compare genes or gene products using semantic similarity measures (SSM). Continuous data growth and diversity in data characteristics comprise what is called big data; current biological SSMs cannot handle big data. Therefore, these measures need the ability to control the size of big data. We used parallel and distributed processing by splitting data into multiple partitions and applied SSM measures to each partition; this approach helped manage big data scalability and computational problems. Our solution involves three steps: split gene ontology (GO), data clustering, and semantic similarity calculation. To test this method, split GO and data clustering algorithms were defined and assessed for performance in the first two steps. Three of the best SSMs in biology [Resnik, Shortest Semantic Differentiation Distance (SSDD), and SORA] are enhanced by introducing threaded parallel processing, which is used in the third step. Our results demonstrate that introducing threads in SSMs reduced the time of calculating semantic similarity between gene pairs and improved performance of the three SSMs. Average time was reduced by 24.51% for Resnik, 22.93%, for SSDD, and 33.68% for SORA. Total time was reduced by 8.88% for Resnik, 23.14% for SSDD, and 39.27% for SORA. Using these threaded measures in the distributed system, combined with using split GO and data clustering algorithms to split input data based on their similarity, reduced the average time more than did the approach of equally dividing input data. Time reduction increased with increasing number of splits. Time reduction percentage was 24.1%, 39.2%, and 66.6% for Threaded SSDD; 33.0%, 78.2%, and 93.1% for Threaded SORA in the case of 2, 3, and 4 slaves, respectively; and 92.04% for Threaded Resnik in the case of four slaves.

Measures of co-expression for improved function prediction of long non-coding RNAs

BACKGROUND

Almost 16,000 human long non-coding RNA (lncRNA) genes have been identified in the GENCODE project. However, the function of most of them remains to be discovered. The function of lncRNAs and other novel genes can be predicted by identifying significantly enriched annotation terms in already annotated genes that are co-expressed with the lncRNAs. However, such approaches are sensitive to the methods that are used to estimate the level of co-expression.

RESULTS

We have tested and compared two well-known statistical metrics (Pearson and Spearman) and two geometrical metrics (Sobolev and Fisher) for identification of the co-expressed genes, using experimental expression data across 19 normal human tissues. We have also used a benchmarking approach based on semantic similarity to evaluate how well these methods are able to predict annotation terms, using a well-annotated set of protein-coding genes.

CONCLUSION

This work shows that geometrical metrics, in particular in combination with the statistical metrics, will predict annotation terms more efficiently than traditional approaches. Tests on selected lncRNAs confirm that it is possible to predict the function of these genes given a reliable set of expression data. The software used for this investigation is freely available.

DDR: efficient computational method to predict drug-target interactions using graph mining and machine learning approaches

Motivation

Finding computationally drug–target interactions (DTIs) is a convenient strategy to identify new DTIs at low cost with reasonable accuracy. However, the current DTI prediction methods suffer the high false positive prediction rate.

Results

We developed DDR, a novel method that improves the DTI prediction accuracy. DDR is based on the use of a heterogeneous graph that contains known DTIs with multiple similarities between drugs and multiple similarities between target proteins. DDR applies non-linear similarity fusion method to combine different similarities. Before fusion, DDR performs a pre-processing step where a subset of similarities is selected in a heuristic process to obtain an optimized combination of similarities. Then, DDR applies a random forest model using different graph-based features extracted from the DTI heterogeneous graph. Using 5-repeats of 10-fold cross-validation, three testing setups, and the weighted average of area under the precision-recall curve (AUPR) scores, we show that DDR significantly reduces the AUPR score error relative to the next best start-of-the-art method for predicting DTIs by 31% when the drugs are new, by 23% when targets are new and by 34% when the drugs and the targets are known but not all DTIs between them are not known. Using independent sources of evidence, we verify as correct 22 out of the top 25 DDR novel predictions. This suggests that DDR can be used as an efficient method to identify correct DTIs.

Availability and implementation

The data and code are provided at https://bitbucket.org/RSO24/ddr/.

Supplementary information

Supplementary data are available at Bioinformatics online.

Gene Ontology Enrichment Improves Performances of Functional Similarity of Genes

There exists a plethora of measures to evaluate functional similarity (FS) between genes, which is a widely used in many bioinformatics applications including detecting molecular pathways, identifying co-expressed genes, predicting protein-protein interactions, and prioritization of disease genes. Measures of FS between genes are mostly derived from Information Contents (IC) of Gene Ontology (GO) terms annotating the genes. However, existing measures evaluating IC of terms based either on the representations of terms in the annotating corpus or on the knowledge embedded in the GO hierarchy do not consider the enrichment of GO terms by the querying pair of genes. The enrichment of a GO term by a pair of gene is dependent on whether the term is annotated by one gene (i.e., partial annotation) or by both genes (i.e. complete annotation) in the pair. In this paper, we propose a method that incorporate enrichment of GO terms by a gene pair in computing their FS and show that GO enrichment improves the performances of 46 existing FS measures in the prediction of sequence homologies, gene expression correlations, protein-protein interactions, and disease associated genes.

An improved approach to infer protein-protein interaction based on a hierarchical vector space model

BACKGROUND

Comparing and classifying functions of gene products are important in today's biomedical research. The semantic similarity derived from the Gene Ontology (GO) annotation has been regarded as one of the most widely used indicators for protein interaction. Among the various approaches proposed, those based on the vector space model are relatively simple, but their effectiveness is far from satisfying.

RESULTS

We propose a Hierarchical Vector Space Model (HVSM) for computing semantic similarity between different genes or their products, which enhances the basic vector space model by introducing the relation between GO terms. Besides the directly annotated terms, HVSM also takes their ancestors and descendants related by "is_a" and "part_of" relations into account. Moreover, HVSM introduces the concept of a Certainty Factor to calibrate the semantic similarity based on the number of terms annotated to genes. To assess the performance of our method, we applied HVSM to Homo sapiens and Saccharomyces cerevisiae protein-protein interaction datasets. Compared with TCSS, Resnik, and other classic similarity measures, HVSM achieved significant improvement for distinguishing positive from negative protein interactions. We also tested its correlation with sequence, EC, and Pfam similarity using online tool CESSM.

CONCLUSIONS

HVSM showed an improvement of up to 4% compared to TCSS, 8% compared to IntelliGO, 12% compared to basic VSM, 6% compared to Resnik, 8% compared to Lin, 11% compared to Jiang, 8% compared to Schlicker, and 11% compared to SimGIC using AUC scores. CESSM test showed HVSM was comparable to SimGIC, and superior to all other similarity measures in CESSM as well as TCSS. Supplementary information and the software are available at https://github.com/kejia1215/HVSM .

Towards a semantic approach to numerical tree inference in phylogenetics

Conventional approaches to phylogeny reconstruction require a character analysis step prior to and methodologically separated from a numerical tree inference step. The former results in a character matrix that contains the empirical data analysed in the latter. This separation of steps involves various methodological and conceptual problems (e.g. homology assessment independent of tree inference and character optimization, character dependencies, discounting of alternative homology hypotheses). In morphology, the character analysis step covers the stages of morphological comparative studies, homology assessment and the identification and coding of morphological characters. Unfortunately, only the last stage requires some formalism, whereas the preceding stages are commonly regarded to be pre-rational and intuitive, which is why their reproducibility and analytical accessibility is limited. Here, I introduce a rational for a semantic approach to numerical tree inference that uses sets of semantic instance anatomies as data source instead of character matrices, thereby avoiding the above-mentioned problems. A semantic instance anatomy is an ontology-based description of the anatomical organization of a specimen in the form of a semantic graph. The semantic approach to numerical tree inference combines and integrates the steps of character analysis and numerical tree inference and makes both analytically accessible and communicable. Before outlining first steps for a research programme dedicated to the semantic approach to numerical tree inference, I discuss in detail the methodological, conceptual, and computational challenges and requirements that first have to be dealt with before adequate algorithms can be developed.

Refine gene functional similarity network based on interaction networks

BACKGROUND

In recent years, biological interaction networks have become the basis of some essential study and achieved success in many applications. Some typical networks such as protein-protein interaction networks have already been investigated systematically. However, little work has been available for the construction of gene functional similarity networks so far. In this research, we will try to build a high reliable gene functional similarity network to promote its further application.

RESULTS

Here, we propose a novel method to construct and refine the gene functional similarity network. It mainly contains three steps. First, we establish an integrated gene functional similarity networks based on different functional similarity calculation methods. Then, we construct a referenced gene-gene association network based on the protein-protein interaction networks. At last, we refine the spurious edges in the integrated gene functional similarity network with the help of the referenced gene-gene association network. Experiment results indicate that the refined gene functional similarity network (RGFSN) exhibits a scale-free, small world and modular architecture, with its degrees fit best to power law distribution. In addition, we conduct protein complex prediction experiment for human based on RGFSN and achieve an outstanding result, which implies it has high reliability and wide application significance.

CONCLUSIONS

Our efforts are insightful for constructing and refining gene functional similarity networks, which can be applied to build other high quality biological networks.