A neural network algorithm for semi-supervised node label learning from unbalanced data

Statistical Learning Methods Applicable to Genome-Wide Association Studies on Unbalanced Case-Control Disease Data

Despite the fact that imbalance between case and control groups is prevalent in genome-wide association studies (GWAS), it is often overlooked. This imbalance is getting more significant and urgent as the rapid growth of biobanks and electronic health records have enabled the collection of thousands of phenotypes from large cohorts, in particular for diseases with low prevalence. The unbalanced binary traits pose serious challenges to traditional statistical methods in terms of both genomic selection and disease prediction. For example, the well-established linear mixed models (LMM) yield inflated type I error rates in the presence of unbalanced case-control ratios. In this article, we review multiple statistical approaches that have been developed to overcome the inaccuracy caused by the unbalanced case-control ratio, with the advantages and limitations of each approach commented. In addition, we also explore the potential for applying several powerful and popular state-of-the-art machine-learning approaches, which have not been applied to the GWAS field yet. This review paves the way for better analysis and understanding of the unbalanced case-control disease data in GWAS.

Epigenetic Element-Based Transcriptome-Wide Association Study Identifies Novel Genes for Bipolar Disorder

Since the bipolar disorder (BD) signals identified by genome-wide association study (GWAS) often reside in the non-coding regions, understanding the biological relevance of these genetic loci has proven to be complicated. Transcriptome-wide association studies (TWAS) providing a powerful approach to identify novel disease risk genes and uncover possible causal genes at loci identified previously by GWAS. However, these methods did not consider the importance of epigenetic regulation in gene expression. Here, we developed a novel epigenetic element-based transcriptome-wide association study (ETWAS) that tested the effects of genetic variants on gene expression levels with the epigenetic features as prior and further mediated the association between predicted expression and BD. We conducted an ETWAS consisting of 20 352 cases and 31 358 controls and identified 44 transcriptome-wide significant hits. We found 14 conditionally independent genes, and 10 genes that did not previously implicate with BD were regarded as novel candidate genes, such as ASB16 in the cerebellar hemisphere (P = 9.29 × 10-8). We demonstrated that several genome-wide significant signals from the BD GWAS driven by genetically regulated expression, and NEK4 explained 90.1% of the GWAS signal. Additionally, ETWAS identified genes could explain heritability beyond that explained by GWAS-associated SNPs (P = 5.60 × 10-66). By querying the SNPs in the final models of identified genes in phenome databases, we identified several phenotypes previously associated with BD, such as schizophrenia and depression. In conclusion, ETWAS is a powerful method, and we identified several novel candidate genes associated with BD.

Multi-resolution visualization and analysis of biomolecular networks through hierarchical community detection and web-based graphical tools

The visual exploration and analysis of biomolecular networks is of paramount importance for identifying hidden and complex interaction patterns among proteins. Although many tools have been proposed for this task, they are mainly focused on the query and visualization of a single protein with its neighborhood. The global exploration of the entire network and the interpretation of its underlying structure still remains difficult, mainly due to the excessively large size of the biomolecular networks. In this paper we propose a novel multi-resolution representation and exploration approach that exploits hierarchical community detection algorithms for the identification of communities occurring in biomolecular networks. The proposed graphical rendering combines two types of nodes (protein and communities) and three types of edges (protein-protein, community-community, protein-community), and displays communities at different resolutions, allowing the user to interactively zoom in and out from different levels of the hierarchy. Links among communities are shown in terms of relationships and functional correlations among the biomolecules they contain. This form of navigation can be also combined by the user with a vertex centric visualization for identifying the communities holding a target biomolecule. Since communities gather limited-size groups of correlated proteins, the visualization and exploration of complex and large networks becomes feasible on off-the-shelf computer machines. The proposed graphical exploration strategies have been implemented and integrated in UNIPred-Web, a web application that we recently introduced for combining the UNIPred algorithm, able to address both integration and protein function prediction in an imbalance-aware fashion, with an easy to use vertex-centric exploration of the integrated network. The tool has been deeply amended from different standpoints, including the prediction core algorithm. Several tests on networks of different size and connectivity have been conducted to show off the vast potential of our methodology; moreover, enrichment analyses have been performed to assess the biological meaningfulness of detected communities. Finally, a CoV-human network has been embedded in the system, and a corresponding case study presented, including the visualization and the prediction of human host proteins that potentially interact with SARS-CoV2 proteins.

Steel Surface Defect Classification Using Deep Residual Neural Network

An automated method for detecting and classifying three classes of surface defects in rolled metal has been developed, which allows for conducting defectoscopy with specified parameters of efficiency and speed. The possibility of using the residual neural networks for classifying defects has been investigated. The classifier based on the ResNet50 neural network is accepted as a basis. The model allows classifying images of flat surfaces with damage of three classes with the general accuracy of 96.91% based on the test data. The use of ResNet50 is shown to provide excellent recognition, high speed, and accuracy, which makes it an effective tool for detecting defects on metal surfaces.

Predicting Functions of Uncharacterized Human Proteins: From Canonical to Proteoforms

Despite tremendous efforts in genomics, transcriptomics, and proteomics communities, there is still no comprehensive data about the exact number of protein-coding genes, translated proteoforms, and their function. In addition, by now, we lack functional annotation for 1193 genes, where expression was confirmed at the proteomic level (uPE1 proteins). We re-analyzed results of AP-MS experiments from the BioPlex 2.0 database to predict functions of uPE1 proteins and their splice forms. By building a protein-protein interaction network for 12 ths. identified proteins encoded by 11 ths. genes, we were able to predict Gene Ontology categories for a total of 387 uPE1 genes. We predicted different functions for canonical and alternatively spliced forms for four uPE1 genes. In total, functional differences were revealed for 62 proteoforms encoded by 31 genes. Based on these results, it can be carefully concluded that the dynamics and versatility of the interactome is ensured by changing the dominant splice form. Overall, we propose that analysis of large-scale AP-MS experiments performed for various cell lines and under various conditions is a key to understanding the full potential of genes role in cellular processes.

A Semisupervised Recurrent Convolutional Attention Model for Human Activity Recognition

Recent years have witnessed the success of deep learning methods in human activity recognition (HAR). The longstanding shortage of labeled activity data inherently calls for a plethora of semisupervised learning methods, and one of the most challenging and common issues with semisupervised learning is the imbalanced distribution of labeled data over classes. Although the problem has long existed in broad real-world HAR applications, it is rarely explored in the literature. In this paper, we propose a semisupervised deep model for imbalanced activity recognition from multimodal wearable sensory data. We aim to address not only the challenges of multimodal sensor data (e.g., interperson variability and interclass similarity) but also the limited labeled data and class-imbalance issues simultaneously. In particular, we propose a pattern-balanced semisupervised framework to extract and preserve diverse latent patterns of activities. Furthermore, we exploit the independence of multi-modalities of sensory data and attentively identify salient regions that are indicative of human activities from inputs by our recurrent convolutional attention networks. Our experimental results demonstrate that the proposed model achieves a competitive performance compared to a multitude of state-of-the-art methods, both semisupervised and supervised ones, with 10% labeled training data. The results also show the robustness of our method over imbalanced, small training data sets.

A distributed semi-supervised learning algorithm based on manifold regularization using wavelet neural network

This paper aims to propose a distributed semi-supervised learning (D-SSL) algorithm to solve D-SSL problems, where training samples are often extremely large-scale and located on distributed nodes over communication networks. Training data of each node consists of labeled and unlabeled samples whose output values or labels are unknown. These nodes communicate in a distributed way, where each node has only access to its own data and can only exchange local information with its neighboring nodes. In some scenarios, these distributed data cannot be processed centrally. As a result, D-SSL problems cannot be centrally solved by using traditional semi-supervised learning (SSL) algorithms. The state-of-the-art D-SSL algorithm, denoted as Distributed Laplacian Regularization Least Square (D-LapRLS), is a kernel based algorithm. It is essential for the D-LapRLS algorithm to estimate the global Euclidian Distance Matrix (EDM) with respect to total samples, which is time-consuming especially when the scale of training data is large. In order to solve D-SSL problems and overcome the common drawback of kernel based D-SSL algorithms, we propose a novel Manifold Regularization (MR) based D-SSL algorithm using Wavelet Neural Network (WNN) and Zero-Gradient-Sum (ZGS) distributed optimization strategy. Accordingly, each node is assigned an individual WNN with the same basis functions. In order to initialize the proposed D-SSL algorithm, we propose a centralized MR based SSL algorithm using WNN. We denote the proposed SSL and D-SSL algorithms as Laplacian WNN (LapWNN) and distributed LapWNN (D-LapWNN), respectively. The D-LapWNN algorithm works in a fully distributed fashion by using ZGS strategy, whose convergence is guaranteed by the Lyapunov method. During the learning process, each node only exchanges local coefficients with its neighbors rather than raw data. It means that the D-LapWNN algorithm is a privacy preserving method. At last, several illustrative simulations are presented to show the efficiency and advantage of the proposed algorithm.

Benchmarking network propagation methods for disease gene identification

In-silico identification of potential target genes for disease is an essential aspect of drug target discovery. Recent studies suggest that successful targets can be found through by leveraging genetic, genomic and protein interaction information. Here, we systematically tested the ability of 12 varied algorithms, based on network propagation, to identify genes that have been targeted by any drug, on gene-disease data from 22 common non-cancerous diseases in OpenTargets. We considered two biological networks, six performance metrics and compared two types of input gene-disease association scores. The impact of the design factors in performance was quantified through additive explanatory models. Standard cross-validation led to over-optimistic performance estimates due to the presence of protein complexes. In order to obtain realistic estimates, we introduced two novel protein complex-aware cross-validation schemes. When seeding biological networks with known drug targets, machine learning and diffusion-based methods found around 2-4 true targets within the top 20 suggestions. Seeding the networks with genes associated to disease by genetics decreased performance below 1 true hit on average. The use of a larger network, although noisier, improved overall performance. We conclude that diffusion-based prioritisers and machine learning applied to diffusion-based features are suited for drug discovery in practice and improve over simpler neighbour-voting methods. We also demonstrate the large impact of choosing an adequate validation strategy and the definition of seed disease genes.

The use of biological network data has proven its effectiveness in many areas from computational biology. Networks consist of nodes, usually genes or proteins, and edges that connect pairs of nodes, representing information such as physical interactions, regulatory roles or co-occurrence. In order to find new candidate nodes for a given biological property, the so-called network propagation algorithms start from the set of known nodes with that property and leverage the connections from the biological network to make predictions. Here, we assess the performance of several network propagation algorithms to find sensible gene targets for 22 common non-cancerous diseases, i.e. those that have been found promising enough to start the clinical trials with any compound. We focus on obtaining performance metrics that reflect a practical scenario in drug development where only a small set of genes can be essayed. We found that the presence of protein complexes biased the performance estimates, leading to over-optimistic conclusions, and introduced two novel strategies to address it. Our results support that network propagation is still a viable approach to find drug targets, but that special care needs to be put on the validation strategy. Algorithms benefitted from the use of a larger -although noisier- network and of direct evidence data, rather than indirect genetic associations to disease.

UNIPred-Web: a web tool for the integration and visualization of biomolecular networks for protein function prediction

Background

One of the main issues in the automated protein function prediction (AFP) problem is the integration of multiple networked data sources. The UNIPred algorithm was thereby proposed to efficiently integrate —in a function-specific fashion— the protein networks by taking into account the imbalance that characterizes protein annotations, and to subsequently predict novel hypotheses about unannotated proteins. UNIPred is publicly available as R code, which might result of limited usage for non-expert users. Moreover, its application requires efforts in the acquisition and preparation of the networks to be integrated. Finally, the UNIPred source code does not handle the visualization of the resulting consensus network, whereas suitable views of the network topology are necessary to explore and interpret existing protein relationships.

Results

We address the aforementioned issues by proposing UNIPred-Web, a user-friendly Web tool for the application of the UNIPred algorithm to a variety of biomolecular networks, already supplied by the system, and for the visualization and exploration of protein networks. We support different organisms and different types of networks —e.g., co-expression, shared domains and physical interaction networks. Users are supported in the different phases of the process, ranging from the selection of the networks and the protein function to be predicted, to the navigation of the integrated network. The system also supports the upload of user-defined protein networks. The vertex-centric and the highly interactive approach of UNIPred-Web allow a narrow exploration of specific proteins, and an interactive analysis of large sub-networks with only a few mouse clicks.

Conclusions

UNIPred-Web offers a practical and intuitive (visual) guidance to biologists interested in gaining insights into protein biomolecular functions. UNIPred-Web provides facilities for the integration of networks, and supplies a framework for the imbalance-aware protein network integration of nine organisms, the prediction of thousands of GO protein functions, and a easy-to-use graphical interface for the visual analysis, navigation and interpretation of the integrated networks and of the functional predictions.

Optimizing gene set annotations combining GO structure and gene expression data

Background

With the rapid accumulation of genomic data, it has become a challenge issue to annotate and interpret these data. As a representative, Gene set enrichment analysis has been widely used to interpret large molecular datasets generated by biological experiments. The result of gene set enrichment analysis heavily relies on the quality and integrity of gene set annotations. Although several methods were developed to annotate gene sets, there is still a lack of high quality annotation methods. Here, we propose a novel method to improve the annotation accuracy through combining the GO structure and gene expression data.

Results

We propose a novel approach for optimizing gene set annotations to get more accurate annotation results. The proposed method filters the inconsistent annotations using GO structure information and probabilistic gene set clusters calculated by a range of cluster sizes over multiple bootstrap resampled datasets. The proposed method is employed to analyze p53 cell lines, colon cancer and breast cancer gene expression data. The experimental results show that the proposed method can filter a number of annotations unrelated to experimental data and increase gene set enrichment power and decrease the inconsistent of annotations.

Conclusions

A novel gene set annotation optimization approach is proposed to improve the quality of gene annotations. Experimental results indicate that the proposed method effectively improves gene set annotation quality based on the GO structure and gene expression data.

A GPU-based algorithm for fast node label learning in large and unbalanced biomolecular networks

BACKGROUND

Several problems in network biology and medicine can be cast into a framework where entities are represented through partially labeled networks, and the aim is inferring the labels (usually binary) of the unlabeled part. Connections represent functional or genetic similarity between entities, while the labellings often are highly unbalanced, that is one class is largely under-represented: for instance in the automated protein function prediction (AFP) for most Gene Ontology terms only few proteins are annotated, or in the disease-gene prioritization problem only few genes are actually known to be involved in the etiology of a given disease. Imbalance-aware approaches to accurately predict node labels in biological networks are thereby required. Furthermore, such methods must be scalable, since input data can be large-sized as, for instance, in the context of multi-species protein networks.

RESULTS

We propose a novel semi-supervised parallel enhancement of COSNET, an imbalance-aware algorithm build on Hopfield neural model recently suggested to solve the AFP problem. By adopting an efficient representation of the graph and assuming a sparse network topology, we empirically show that it can be efficiently applied to networks with millions of nodes. The key strategy to speed up the computations is to partition nodes into independent sets so as to process each set in parallel by exploiting the power of GPU accelerators. This parallel technique ensures the convergence to asymptotically stable attractors, while preserving the asynchronous dynamics of the original model. Detailed experiments on real data and artificial big instances of the problem highlight scalability and efficiency of the proposed method.

CONCLUSIONS

By parallelizing COSNET we achieved on average a speed-up of 180x in solving the AFP problem in the S. cerevisiae, Mus musculus and Homo sapiens organisms, while lowering memory requirements. In addition, to show the potential applicability of the method to huge biomolecular networks, we predicted node labels in artificially generated sparse networks involving hundreds of thousands to millions of nodes.

Predicting Protein Function via Semantic Integration of Multiple Networks

Determining the biological functions of proteins is one of the key challenges in the post-genomic era. The rapidly accumulated large volumes of proteomic and genomic data drives to develop computational models for automatically predicting protein function in large scale. Recent approaches focus on integrating multiple heterogeneous data sources and they often get better results than methods that use single data source alone. In this paper, we investigate how to integrate multiple biological data sources with the biological knowledge, i.e., Gene Ontology (GO), for protein function prediction. We propose a method, called SimNet, to Semantically integrate multiple functional association Networks derived from heterogenous data sources. SimNet firstly utilizes GO annotations of proteins to capture the semantic similarity between proteins and introduces a semantic kernel based on the similarity. Next, SimNet constructs a composite network, obtained as a weighted summation of individual networks, and aligns the network with the kernel to get the weights assigned to individual networks. Then, it applies a network-based classifier on the composite network to predict protein function. Experiment results on heterogenous proteomic data sources of Yeast, Human, Mouse, and Fly show that, SimNet not only achieves better (or comparable) results than other related competitive approaches, but also takes much less time. The Matlab codes of SimNet are available at https://sites.google.com/site/guoxian85/simnet.

UNIPred: Unbalance-Aware Network Integration and Prediction of Protein Functions

The proper integration of multiple sources of data and the unbalance between annotated and unannotated proteins represent two of the main issues of the automated function prediction (AFP) problem. Most of supervised and semisupervised learning algorithms for AFP proposed in literature do not jointly consider these items, with a negative impact on both sensitivity and precision performances, due to the unbalance between annotated and unannotated proteins that characterize the majority of functional classes and to the specific and complementary information content embedded in each available source of data. We propose UNIPred (unbalance-aware network integration and prediction of protein functions), an algorithm that properly combines different biomolecular networks and predicts protein functions using parametric semisupervised neural models. The algorithm explicitly takes into account the unbalance between unannotated and annotated proteins both to construct the integrated network and to predict protein annotations for each functional class. Full-genome and ontology-wide experiments with three eukaryotic model organisms show that the proposed method compares favorably with state-of-the-art learning algorithms for AFP.

Hierarchical ensemble methods for protein function prediction

Protein function prediction is a complex multiclass multilabel classification problem, characterized by multiple issues such as the incompleteness of the available annotations, the integration of multiple sources of high dimensional biomolecular data, the unbalance of several functional classes, and the difficulty of univocally determining negative examples. Moreover, the hierarchical relationships between functional classes that characterize both the Gene Ontology and FunCat taxonomies motivate the development of hierarchy-aware prediction methods that showed significantly better performances than hierarchical-unaware "flat" prediction methods. In this paper, we provide a comprehensive review of hierarchical methods for protein function prediction based on ensembles of learning machines. According to this general approach, a separate learning machine is trained to learn a specific functional term and then the resulting predictions are assembled in a "consensus" ensemble decision, taking into account the hierarchical relationships between classes. The main hierarchical ensemble methods proposed in the literature are discussed in the context of existing computational methods for protein function prediction, highlighting their characteristics, advantages, and limitations. Open problems of this exciting research area of computational biology are finally considered, outlining novel perspectives for future research.

Think globally and solve locally: secondary memory-based network learning for automated multi-species function prediction

Background

Network-based learning algorithms for automated function prediction (AFP) are negatively affected by the limited coverage of experimental data and limited a priori known functional annotations. As a consequence their application to model organisms is often restricted to well characterized biological processes and pathways, and their effectiveness with poorly annotated species is relatively limited. A possible solution to this problem might consist in the construction of big networks including multiple species, but this in turn poses challenging computational problems, due to the scalability limitations of existing algorithms and the main memory requirements induced by the construction of big networks. Distributed computation or the usage of big computers could in principle respond to these issues, but raises further algorithmic problems and require resources not satisfiable with simple off-the-shelf computers.

Results

We propose a novel framework for scalable network-based learning of multi-species protein functions based on both a local implementation of existing algorithms and the adoption of innovative technologies: we solve “locally” the AFP problem, by designing “vertex-centric” implementations of network-based algorithms, but we do not give up thinking “globally” by exploiting the overall topology of the network. This is made possible by the adoption of secondary memory-based technologies that allow the efficient use of the large memory available on disks, thus overcoming the main memory limitations of modern off-the-shelf computers. This approach has been applied to the analysis of a large multi-species network including more than 300 species of bacteria and to a network with more than 200,000 proteins belonging to 13 Eukaryotic species. To our knowledge this is the first work where secondary-memory based network analysis has been applied to multi-species function prediction using biological networks with hundreds of thousands of proteins.

Conclusions

The combination of these algorithmic and technological approaches makes feasible the analysis of large multi-species networks using ordinary computers with limited speed and primary memory, and in perspective could enable the analysis of huge networks (e.g. the whole proteomes available in SwissProt), using well-equipped stand-alone machines.