Discriminating cirRNAs from other lncRNAs using a hierarchical extreme learning machine (H-ELM) algorithm with feature selection

CircRNA identification and feature interpretability analysis

BACKGROUND

Circular RNAs (circRNAs) can regulate microRNA activity and are related to various diseases, such as cancer. Functional research on circRNAs is the focus of scientific research. Accurate identification of circRNAs is important for gaining insight into their functions. Although several circRNA prediction models have been developed, their prediction accuracy is still unsatisfactory. Therefore, providing a more accurate computational framework to predict circRNAs and analyse their looping characteristics is crucial for systematic annotation.

RESULTS

We developed a novel framework, CircDC, for classifying circRNAs from other lncRNAs. CircDC uses four different feature encoding schemes and adopts a multilayer convolutional neural network and bidirectional long short-term memory network to learn high-order feature representation and make circRNA predictions. The results demonstrate that the proposed CircDC model is more accurate than existing models. In addition, an interpretable analysis of the features affecting the model is performed, and the computational framework is applied to the extended application of circRNA identification.

CONCLUSIONS

CircDC is suitable for the prediction of circRNA. The identification of circRNA helps to understand and delve into the related biological processes and functions. Feature importance analysis increases model interpretability and uncovers significant biological properties. The relevant code and data in this article can be accessed for free at https://github.com/nmt315320/CircDC.git .

A review on extreme learning machine

Extreme learning machine (ELM) is a training algorithm for single hidden layer feedforward neural network (SLFN), which converges much faster than traditional methods and yields promising performance. In this paper, we hope to present a comprehensive review on ELM. Firstly, we will focus on the theoretical analysis including universal approximation theory and generalization. Then, the various improvements are listed, which help ELM works better in terms of stability, efficiency, and accuracy. Because of its outstanding performance, ELM has been successfully applied in many real-time learning tasks for classification, clustering, and regression. Besides, we report the applications of ELM in medical imaging: MRI, CT, and mammogram. The controversies of ELM were also discussed in this paper. We aim to report these advances and find some future perspectives.

Attention-Based Deep Multiple-Instance Learning for Classifying Circular RNA and Other Long Non-Coding RNA

Circular RNA (circRNA) is a distinguishable circular formed long non-coding RNA (lncRNA), which has specific roles in transcriptional regulation, multiple biological processes. The identification of circRNA from other lncRNA is necessary for relevant research. In this study, we designed attention-based multi-instance learning (MIL) network architecture fed with a raw sequence, to learn the sparse features of RNA sequences and to accomplish the circRNAs identification task. The model outperformed the state-of-art models. Moreover, following the validation of the attention mechanism effectiveness by the handwritten digit dataset, the key sequence loci underlying circRNA’s recognition were obtained based on the corresponding attention score. Then, motif enrichment analysis identified some of the key motifs for circRNA formation. In conclusion, we designed deep learning network architecture suitable for learning gene sequences with sparse features and implemented it for the circRNA identification task, and the model has strong representation capability in the indication of some key loci.

Characterizing viral circRNAs and their application in identifying circRNAs in viruses

Circular RNAs (circRNAs) are non-coding RNAs with a special circular structure produced formed by the reverse splicing mechanism, which play an important role in a variety of biological activities. Viruses can encode circRNA, and viral circRNAs have been found in multiple single-stranded and double-stranded viruses. However, the characteristics and functions of viral circRNAs remain unknown. Sequence alignment showed that viral circRNAs are less conserved than circRNAs in animal, indicating that the viral circRNAs may evolve rapidly. Through the analysis of the sequence characteristics of viral circRNAs and circRNAs in animal, it was found that viral circRNAs and animals circRNAs are similar in nucleic acid composition, but have obvious differences in secondary structure and autocorrelation characteristics. Based on these characteristics of viral circRNAs, machine learning algorithms were employed to construct a prediction model to identify viral circRNA. Additionally, analysis of the interaction between viral circRNA and miRNAs showed that viral circRNA is expected to interact with 518 human miRNAs, and preliminary analysis of the role of viral circRNA. And it has been also found that viral circRNAs may be involved in many KEGG pathways related to nervous system and cancer. We curated an online server, and the data and code are available: http://server.malab.cn/viral-CircRNA/.

Advances in Computational Methodologies for Classification and Sub-Cellular Locality Prediction of Non-Coding RNAs

Apart from protein-coding Ribonucleic acids (RNAs), there exists a variety of non-coding RNAs (ncRNAs) which regulate complex cellular and molecular processes. High-throughput sequencing technologies and bioinformatics approaches have largely promoted the exploration of ncRNAs which revealed their crucial roles in gene regulation, miRNA binding, protein interactions, and splicing. Furthermore, ncRNAs are involved in the development of complicated diseases like cancer. Categorization of ncRNAs is essential to understand the mechanisms of diseases and to develop effective treatments. Sub-cellular localization information of ncRNAs demystifies diverse functionalities of ncRNAs. To date, several computational methodologies have been proposed to precisely identify the class as well as sub-cellular localization patterns of RNAs). This paper discusses different types of ncRNAs, reviews computational approaches proposed in the last 10 years to distinguish coding-RNA from ncRNA, to identify sub-types of ncRNAs such as piwi-associated RNA, micro RNA, long ncRNA, and circular RNA, and to determine sub-cellular localization of distinct ncRNAs and RNAs. Furthermore, it summarizes diverse ncRNA classification and sub-cellular localization determination datasets along with benchmark performance to aid the development and evaluation of novel computational methodologies. It identifies research gaps, heterogeneity, and challenges in the development of computational approaches for RNA sequence analysis. We consider that our expert analysis will assist Artificial Intelligence researchers with knowing state-of-the-art performance, model selection for various tasks on one platform, dominantly used sequence descriptors, neural architectures, and interpreting inter-species and intra-species performance deviation.

JEDI: circular RNA prediction based on junction encoders and deep interaction among splice sites

Motivation

Circular RNA (circRNA) is a novel class of long non-coding RNAs that have been broadly discovered in the eukaryotic transcriptome. The circular structure arises from a non-canonical splicing process, where the donor site backspliced to an upstream acceptor site. These circRNA sequences are conserved across species. More importantly, rising evidence suggests their vital roles in gene regulation and association with diseases. As the fundamental effort toward elucidating their functions and mechanisms, several computational methods have been proposed to predict the circular structure from the primary sequence. Recently, advanced computational methods leverage deep learning to capture the relevant patterns from RNA sequences and model their interactions to facilitate the prediction. However, these methods fail to fully explore positional information of splice junctions and their deep interaction.

Results

We present a robust end-to-end framework, Junction Encoder with Deep Interaction (JEDI), for circRNA prediction using only nucleotide sequences. JEDI first leverages the attention mechanism to encode each junction site based on deep bidirectional recurrent neural networks and then presents the novel cross-attention layer to model deep interaction among these sites for backsplicing. Finally, JEDI can not only predict circRNAs but also interpret relationships among splice sites to discover backsplicing hotspots within a gene region. Experiments demonstrate JEDI significantly outperforms state-of-the-art approaches in circRNA prediction on both isoform level and gene level. Moreover, JEDI also shows promising results on zero-shot backsplicing discovery, where none of the existing approaches can achieve.

Availability and implementation

The implementation of our framework is available at https://github.com/hallogameboy/JEDI.

Supplementary information

Supplementary data are available at Bioinformatics online.

Current research on circular RNAs and their potential clinical implications in breast cancer

Breast cancer (BC) is one of the most common cancers and the leading causes of death among women worldwide, and its morbidity rate is growing. Discovery of novel biomarkers is necessary for early BC detection, treatment, and prognostication. Circular RNAs (circRNAs), a novel type of endogenous non-coding RNAs with covalently closed continuous loops, have been found to have a crucial role in tumorigenesis. Studies have demonstrated that circRNAs are aberrantly expressed in the tumor tissues and plasma of patients with BC, and they modulate gene expression affecting the proliferation, metastasis, and chemoresistance of BC by specifically binding and regulating the expression of microRNAs (miRNAs). Therefore, circRNAs can be used as novel potential diagnostic and prognostic markers, and therapeutic targets for BC. This article summarizes the properties, functions, and regulatory mechanisms of circRNAs, particularly current research on their association with BC proliferation, metastasis, and chemoresistance.

Identification of multiple RNAs using feature fusion

Detection of novel transcripts with deep sequencing has increased the demand for computational algorithms as their identification and validation using in vivo techniques is time-consuming, costly and unreliable. Most of these discovered transcripts belong to non-coding RNAs, a large group known for their diverse functional roles but lacks the common taxonomy. Thus, upon the identification of the absence of coding potential in them, it is crucial to recognize their prime functional category. To address this heterogeneity issue, we divide the ncRNAs into three classes and present RNA classifier (RNAC) that categorizes the RNAs into coding, housekeeping, small non-coding and long non-coding classes. RNAC utilizes the alignment-based genomic descriptors to extract statistical, local binary patterns and histogram features and fuse them to construct the classification models with extreme gradient boosting. The experiments are performed on four species, and the performance is assessed on multiclass and conventional binary classification (coding versus no-coding) problems. The proposed approach achieved >93% accuracy on both classification problems and also outperformed other well-known existing methods in coding potential prediction. This validates the usefulness of feature fusion for improved performance on both types of classification problems. Hence, RNAC is a valuable tool for the accurate identification of multiple RNAs .

Feature extraction approaches for biological sequences: a comparative study of mathematical features

As consequence of the various genomic sequencing projects, an increasing volume of biological sequence data is being produced. Although machine learning algorithms have been successfully applied to a large number of genomic sequence-related problems, the results are largely affected by the type and number of features extracted. This effect has motivated new algorithms and pipeline proposals, mainly involving feature extraction problems, in which extracting significant discriminatory information from a biological set is challenging. Considering this, our work proposes a new study of feature extraction approaches based on mathematical features (numerical mapping with Fourier, entropy and complex networks). As a case study, we analyze long non-coding RNA sequences. Moreover, we separated this work into three studies. First, we assessed our proposal with the most addressed problem in our review, e.g. lncRNA and mRNA; second, we also validate the mathematical features in different classification problems, to predict the class of lncRNA, e.g. circular RNAs sequences; third, we analyze its robustness in scenarios with imbalanced data. The experimental results demonstrated three main contributions: first, an in-depth study of several mathematical features; second, a new feature extraction pipeline; and third, its high performance and robustness for distinct RNA sequence classification. Availability:  https://github.com/Bonidia/FeatureExtraction_BiologicalSequences.

The Methylation Pattern for Knee and Hip Osteoarthritis

Osteoarthritis is one of the most prevalent chronic joint diseases for middle-aged and elderly people. But in recent years, the number of young people suffering from the disease increases quickly. It is known that osteoarthritis is a common degenerative disease caused by the combination and interaction of many factors such as natural and environmental factors. DNA methylations reflect the effects of environmental factors. Several researches on DNA methylation at specific genes in OA cartilage indicated the great potential roles of DNA methylation in OA. To systematically investigate the methylation pattern in knee and hip osteoarthritis, we analyzed the methylation profiles in cartilage of 16 OA hip samples, 19 control hip samples and 62 OA knee samples. 12 discriminative methylation sites were identified using advanced minimal Redundancy Maximal Relevance (mRMR) and Incremental Feature Selection (IFS) methods. The SVM classifier of these 12 methylation sites from genes like MEIS1, GABRG3, RXRA, and EN1, can perfectly classify the OA hip samples, control hip samples and OA knee samples evaluated with LOOCV (Leave-One Out-Cross Validation). These 12 methylation sites can not only serve as biomarker, but also provide underlying mechanism of OA.

RNMFMDA: A Microbe-Disease Association Identification Method Based on Reliable Negative Sample Selection and Logistic Matrix Factorization With Neighborhood Regularization

Microbes with abnormal levels have important impacts on the formation and development of various complex diseases. Identifying possible Microbe-Disease Associations (MDAs) helps to understand the mechanisms of complex diseases. However, experimental methods for MDA identification are costly and time-consuming. In this study, a new computational model, RNMFMDA, was developed to find possible MDAs. RNMFMDA contains two main processes. First, Reliable Negative MDA samples were selected based on Positive-Unlabeled (PU) learning and random walk with restart on the heterogeneous microbe-disease network. Second, Logistic Matrix Factorization with Neighborhood Regularization (LMFNR) was developed to compute the association probabilities for all microbe-disease pairs. To evaluate the performance of the proposed RNMFMDA method, we compared RNMFMDA with five state-of-the-art MDA prediction methods based on five-fold cross-validations on microbes, diseases, and MDAs. As a result, RNMFMDA obtained the best AUCs of 0.6332, 0.8669, and 0.9081, respectively for the three five-fold cross validations, significantly outperforming other models. The promising prediction performance may be attributed to the following three features: highly quality negative MDA sample selection, LMFNR-based MDA prediction model, and various biological information integration. In addition, a few predicted microbe-disease pairs with high association scores are worthy of further experimental validation.

The Key Genes for Perineural Invasion in Pancreatic Ductal Adenocarcinoma Identified With Monte-Carlo Feature Selection Method

Background

Pancreatic ductal adenocarcinoma (PDAC) is the most aggressive form of pancreatic cancer. Its 5-year survival rate is only 3–5%. Perineural invasion (PNI) is a process of cancer cells invading the surrounding nerves and perineural spaces. It is considered to be associated with the poor prognosis of PDAC. About 90% of pancreatic cancer patients have PNI. The high incidence of PNI in pancreatic cancer limits radical resection and promotes local recurrence, which negatively affects life quality and survival time of the patients with pancreatic cancer.

Objectives

To investigate the mechanism of PNI in pancreatic cancer, we analyzed the gene expression profiles of tumors and adjacent tissues from 50 PDAC patients which included 28 patients with perineural invasion and 22 patients without perineural invasion.

Method

Using Monte-Carlo feature selection and Incremental Feature Selection (IFS) method, we identified 26 key features within which 15 features were from tumor tissues and 11 features were from adjacent tissues.

Results

Our results suggested that not only the tumor tissue, but also the adjacent tissue, was informative for perineural invasion prediction. The SVM classifier based on these 26 key features can predict perineural invasion accurately, with a high accuracy of 0.94 evaluated with leave-one-out cross validation (LOOCV).

Conclusion

The in-depth biological analysis of key feature genes, such as TNFRSF14, XPO1, and ATF3, shed light on the understanding of perineural invasion in pancreatic ductal adenocarcinoma.

New Computational Tool Based on Machine-learning Algorithms for the Identification of Rhinovirus Infection-Related Genes

BACKGROUND

Human rhinovirus has different identified serotypes and is the most common cause of cold in humans. To date, many genes have been discovered to be related to rhinovirus infection. However, the pathogenic mechanism of rhinovirus is difficult to elucidate through experimental approaches due to the high cost and consuming time.

METHODS AND RESULTS

In this study, we presented a novel approach that relies on machine-learning algorithms and identified two genes OTOF and SOCS1. The expression levels of these genes in the blood samples can be used to accurately distinguish virus-infected and non-infected individuals.

CONCLUSION

Our findings suggest the crucial roles of these two genes in rhinovirus infection and the robustness of the computational tool in dissecting pathogenic mechanisms.

circDeep: deep learning approach for circular RNA classification from other long non-coding RNA

MOTIVATION

Over the past two decades, a circular form of RNA (circular RNA), produced through alternative splicing, has become the focus of scientific studies due to its major role as a microRNA (miRNA) activity modulator and its association with various diseases including cancer. Therefore, the detection of circular RNAs is vital to understanding their biogenesis and purpose. Prediction of circular RNA can be achieved in three steps: distinguishing non-coding RNAs from protein coding gene transcripts, separating short and long non-coding RNAs and predicting circular RNAs from other long non-coding RNAs (lncRNAs). However, the available tools are less than 80 percent accurate for distinguishing circular RNAs from other lncRNAs due to difficulty of classification. Therefore, the availability of a more accurate and fast machine learning method for the identification of circular RNAs, which considers the specific features of circular RNA, is essential to the development of systematic annotation.

RESULTS

Here we present an End-to-End deep learning framework, circDeep, to classify circular RNA from other lncRNA. circDeep fuses an RCM descriptor, ACNN-BLSTM sequence descriptor and a conservation descriptor into high level abstraction descriptors, where the shared representations across different modalities are integrated. The experiments show that circDeep is not only faster than existing tools but also performs at an unprecedented level of accuracy by achieving a 12 percent increase in accuracy over the other tools.

AVAILABILITY AND IMPLEMENTATION

https://github.com/UofLBioinformatics/circDeep.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Identifying Circular RNA and Predicting Its Regulatory Interactions by Machine Learning

Circular RNA (circRNA) is a closed long non-coding RNA (lncRNA) formed by covalently closed loops through back-splicing. Emerging evidence indicates that circRNA can influence cellular physiology through various molecular mechanisms. Thus, accurate circRNA identification and prediction of its regulatory information are critical for understanding its biogenesis. Although several computational tools based on machine learning have been proposed for circRNA identification, the prediction accuracy remains to be improved. Here, first we present circLGB, a machine learning-based framework to discriminate circRNA from other lncRNAs. circLGB integrates commonly used sequence-derived features and three new features containing adenosine to inosine (A-to-I) deamination, A-to-I density and the internal ribosome entry site. circLGB categorizes circRNAs by utilizing a LightGBM classifier with feature selection. Second, we introduce circMRT, an ensemble machine learning framework to systematically predict the regulatory information for circRNA, including their interactions with microRNA, the RNA binding protein, and transcriptional regulation. Feature sets including sequence-based features, graph features, genome context, and regulatory information features were modeled in circMRT. Experiments on public and our constructed datasets show that the proposed algorithms outperform the available state-of-the-art methods. circLGB is available at http://www.circlgb.com. Source codes are available at https://github.com/Peppags/circLGB-circMRT.

Identification of Post-myocardial Infarction Blood Expression Signatures Using Multiple Feature Selection Strategies

Myocardial infarction (MI) is a type of serious heart attack in which the blood flow to the heart is suddenly interrupted, resulting in injury to the heart muscles due to a lack of oxygen supply. Although clinical diagnosis methods can be used to identify the occurrence of MI, using the changes of molecular markers or characteristic molecules in blood to characterize the early phase and later trend of MI will help us choose a more reasonable treatment plan. Previously, comparative transcriptome studies focused on finding differentially expressed genes between MI patients and healthy people. However, signature molecules altered in different phases of MI have not been well excavated. We developed a set of computational approaches integrating multiple machine learning algorithms, including Monte Carlo feature selection (MCFS), incremental feature selection (IFS), and support vector machine (SVM), to identify gene expression characteristics on different phases of MI. 134 genes were determined to serve as features for building optimal SVM classifiers to distinguish acute MI and post-MI. Subsequently, functional enrichment analyses followed by protein-protein interaction analysis on 134 genes identified several hub genes (IL1R1, TLR2, and TLR4) associated with progression of MI, which can be used as new diagnostic molecules for MI.

Measuring Performance Metrics of Machine Learning Algorithms for Detecting and Classifying Transposable Elements

Because of the promising results obtained by machine learning (ML) approaches in several fields, every day is more common, the utilization of ML to solve problems in bioinformatics. In genomics, a current issue is to detect and classify transposable elements (TEs) because of the tedious tasks involved in bioinformatics methods. Thus, ML was recently evaluated for TE datasets, demonstrating better results than bioinformatics applications. A crucial step for ML approaches is the selection of metrics that measure the realistic performance of algorithms. Each metric has specific characteristics and measures properties that may be different from the predicted results. Although the most commonly used way to compare measures is by using empirical analysis, a non-result-based methodology has been proposed, called measure invariance properties. These properties are calculated on the basis of whether a given measure changes its value under certain modifications in the confusion matrix, giving comparative parameters independent of the datasets. Measure invariance properties make metrics more or less informative, particularly on unbalanced, monomodal, or multimodal negative class datasets and for real or simulated datasets. Although several studies applied ML to detect and classify TEs, there are no works evaluating performance metrics in TE tasks. Here, we analyzed 26 different metrics utilized in binary, multiclass, and hierarchical classifications, through bibliographic sources, and their invariance properties. Then, we corroborated our findings utilizing freely available TE datasets and commonly used ML algorithms. Based on our analysis, the most suitable metrics for TE tasks must be stable, even using highly unbalanced datasets, multimodal negative class, and training datasets with errors or outliers. Based on these parameters, we conclude that the F1-score and the area under the precision-recall curve are the most informative metrics since they are calculated based on other metrics, providing insight into the development of an ML application.

CirRNAPL: A web server for the identification of circRNA based on extreme learning machine

Circular RNA (circRNA) plays an important role in the development of diseases, and it provides a novel idea for drug development. Accurate identification of circRNAs is important for a deeper understanding of their functions. In this study, we developed a new classifier, CirRNAPL, which extracts the features of nucleic acid composition and structure of the circRNA sequence and optimizes the extreme learning machine based on the particle swarm optimization algorithm. We compared CirRNAPL with existing methods, including blast, on three datasets and found CirRNAPL significantly improved the identification accuracy for the three datasets, with accuracies of 0.815, 0.802, and 0.782, respectively. Additionally, we performed sequence alignment on 564 sequences of the independent detection set of the third data set and analyzed the expression level of circRNAs. Results showed the expression level of the sequence is positively correlated with the abundance. A user-friendly CirRNAPL web server is freely available at http://server.malab.cn/CirRNAPL/.

The Functional Effects of Key Driver KRAS Mutations on Gene Expression in Lung Cancer

Lung cancer is a common malignant cancer. Kirsten rat sarcoma oncogene (KRAS) mutations have been considered as a key driver for lung cancers. KRAS p.G12C mutations were most predominant in NSCLC which was comprised about 11–16% of lung adenocarcinomas (p.G12C accounts for 45–50% of mutant KRAS). But it is still not clear how the KRAS mutation triggers lung cancers. To study the molecular mechanisms of KRAS mutation in lung cancer. We analyzed the gene expression profiles of 156 KRAS mutation samples and other negative samples with two stage feature selection approach: (1) minimal Redundancy Maximal Relevance (mRMR) and (2) Incremental Feature Selection (IFS). At last, 41 predictive genes for KRAS mutation were identified and a KRAS mutation predictor was constructed. Its leave one out cross validation MCC was 0.879. Our results were helpful for understanding the roles of KRAS mutation in lung cancer.

A systematic review of the application of machine learning in the detection and classification of transposable elements

Background

Transposable elements (TEs) constitute the most common repeated sequences in eukaryotic genomes. Recent studies demonstrated their deep impact on species diversity, adaptation to the environment and diseases. Although there are many conventional bioinformatics algorithms for detecting and classifying TEs, none have achieved reliable results on different types of TEs. Machine learning (ML) techniques can automatically extract hidden patterns and novel information from labeled or non-labeled data and have been applied to solving several scientific problems.

Methodology

We followed the Systematic Literature Review (SLR) process, applying the six stages of the review protocol from it, but added a previous stage, which aims to detect the need for a review. Then search equations were formulated and executed in several literature databases. Relevant publications were scanned and used to extract evidence to answer research questions.

Results

Several ML approaches have already been tested on other bioinformatics problems with promising results, yet there are few algorithms and architectures available in literature focused specifically on TEs, despite representing the majority of the nuclear DNA of many organisms. Only 35 articles were found and categorized as relevant in TE or related fields.

Conclusions

ML is a powerful tool that can be used to address many problems. Although ML techniques have been used widely in other biological tasks, their utilization in TE analyses is still limited. Following the SLR, it was possible to notice that the use of ML for TE analyses (detection and classification) is an open problem, and this new field of research is growing in interest.

Identifying Microbiota Signature and Functional Rules Associated With Bacterial Subtypes in Human Intestine

Gut microbiomes are integral microflora located in the human intestine with particular symbiosis. Among all microorganisms in the human intestine, bacteria are the most significant subgroup that contains many unique and functional species. The distribution patterns of bacteria in the human intestine not only reflect the different microenvironments in different sections of the intestine but also indicate that bacteria may have unique biological functions corresponding to their proper regions of the intestine. However, describing the functional differences between the bacterial subgroups and their distributions in different individuals is difficult using traditional computational approaches. Here, we first attempted to introduce four effective sets of bacterial features from independent databases. We then presented a novel computational approach to identify potential distinctive features among bacterial subgroups based on a systematic dataset on the gut microbiome from approximately 1,500 human gut bacterial strains. We also established a group of quantitative rules for explaining such distinctions. Results may reveal the microstructural characteristics of the intestinal flora and deepen our understanding on the regulatory role of bacterial subgroups in the human intestine.

Prediction of Citrullination Sites on the Basis of mRMR Method and SNN

BACKGROUND

Citrullination, an important post-translational modification of proteins, alters the molecular weight and electrostatic charge of the protein side chains. Citrulline, in protein sequences, is catalyzed by a class of Peptidyl Arginine Deiminases (PADs). Dependent on Ca2+, PADs include five isozymes: PAD 1, 2, 3, 4/5, and 6. Citrullinated proteins have been identified in many biological and pathological processes. Among them, abnormal protein citrullination modification can lead to serious human diseases, including multiple sclerosis and rheumatoid arthritis.

OBJECTIVE

It is important to identify the citrullination sites in protein sequences. The accurate identification of citrullination sites may contribute to the studies on the molecular functions and pathological mechanisms of related diseases.

METHODS AND RESULTS

In this study, after an encoded training set (containing 116 positive and 348 negative samples) into the feature matrix, the mRMR method was used to analyze the 941- dimensional features which were sorted on the basis of their importance. Then, a predictive model based on a self-normalizing neural network (SNN) was proposed to predict the citrullination sites in protein sequences. Incremental Feature Selection (IFS) and 10-fold cross-validation were used as the model evaluation method. Three classical machine learning models, namely random forest, support vector machine, and k-nearest neighbor algorithm, were selected and compared with the SNN prediction model using the same evaluation methods. SNN may be the best tool for citrullination site prediction. The maximum value of the Matthews Correlation Coefficient (MCC) reached 0.672404 on the basis of the optimal classifier of SNN.

CONCLUSION

The results showed that the SNN-based prediction methods performed better when evaluated by some common metrics, such as MCC, accuracy, and F1-Measure. SNN prediction model also achieved a better balance in the classification and recognition of positive and negative samples from datasets compared with the other three models.

Identifying Methylation Pattern and Genes Associated with Breast Cancer Subtypes

Breast cancer is regarded worldwide as a severe human disease. Various genetic variations, including hereditary and somatic mutations, contribute to the initiation and progression of this disease. The diagnostic parameters of breast cancer are not limited to the conventional protein content and can include newly discovered genetic variants and even genetic modification patterns such as methylation and microRNA. In addition, breast cancer detection extends to detailed breast cancer stratifications to provide subtype-specific indications for further personalized treatment. One genome-wide expression–methylation quantitative trait loci analysis confirmed that different breast cancer subtypes have various methylation patterns. However, recognizing clinically applied (methylation) biomarkers is difficult due to the large number of differentially methylated genes. In this study, we attempted to re-screen a small group of functional biomarkers for the identification and distinction of different breast cancer subtypes with advanced machine learning methods. The findings may contribute to biomarker identification for different breast cancer subtypes and provide a new perspective for differential pathogenesis in breast cancer subtypes.

Analysis of Expression Pattern of snoRNAs in Different Cancer Types with Machine Learning Algorithms

Small nucleolar RNAs (snoRNAs) are a new type of functional small RNAs involved in the chemical modifications of rRNAs, tRNAs, and small nuclear RNAs. It is reported that they play important roles in tumorigenesis via various regulatory modes. snoRNAs can both participate in the regulation of methylation and pseudouridylation and regulate the expression pattern of their host genes. This research investigated the expression pattern of snoRNAs in eight major cancer types in TCGA via several machine learning algorithms. The expression levels of snoRNAs were first analyzed by a powerful feature selection method, Monte Carlo feature selection (MCFS). A feature list and some informative features were accessed. Then, the incremental feature selection (IFS) was applied to the feature list to extract optimal features/snoRNAs, which can make the support vector machine (SVM) yield best performance. The discriminative snoRNAs included HBII-52-14, HBII-336, SNORD123, HBII-85-29, HBII-420, U3, HBI-43, SNORD116, SNORA73B, SCARNA4, HBII-85-20, etc., on which the SVM can provide a Matthew’s correlation coefficient (MCC) of 0.881 for predicting these eight cancer types. On the other hand, the informative features were fed into the Johnson reducer and repeated incremental pruning to produce error reduction (RIPPER) algorithms to generate classification rules, which can clearly show different snoRNAs expression patterns in different cancer types. The analysis results indicated that extracted discriminative snoRNAs can be important for identifying cancer samples in different types and the expression pattern of snoRNAs in different cancer types can be partly uncovered by quantitative recognition rules.

Computational Method for the Identification of Molecular Metabolites Involved in Cereal Hull Color Variations

Background:

Cereal hull color is an important quality specification characteristic. Many studies were conducted to identify genetic changes underlying cereal hull color diversity. However, these studies mainly focused on the gene level. Recent studies have suggested that metabolomics can accurately reflect the integrated and real-time cell processes that contribute to the formation of different cereal colors.

Methods:

In this study, we exploited published metabolomics databases and applied several advanced computational methods, such as minimum redundancy maximum relevance (mRMR), incremental forward search (IFS), random forest (RF) to investigate cereal hull color at the metabolic level. First, the mRMR was applied to analyze cereal hull samples represented by metabolite features, yielding a feature list. Then, the IFS and RF were used to test several feature sets, constructed according to the aforementioned feature list. Finally, the optimal feature sets and RF classifier were accessed based on the testing results.

Results and Conclusion:

A total of 158 key metabolites were found to be useful in distinguishing white cereal hulls from colorful cereal hulls. A prediction model constructed with these metabolites and a random forest algorithm generated a high Matthews coefficient correlation value of 0.701. Furthermore, 24 of these metabolites were previously found to be relevant to cereal color. Our study can provide new insights into the molecular basis of cereal hull color formation.

Machine Learning and Integrative Analysis of Biomedical Big Data

Recent developments in high-throughput technologies have accelerated the accumulation of massive amounts of omics data from multiple sources: genome, epigenome, transcriptome, proteome, metabolome, etc. Traditionally, data from each source (e.g., genome) is analyzed in isolation using statistical and machine learning (ML) methods. Integrative analysis of multi-omics and clinical data is key to new biomedical discoveries and advancements in precision medicine. However, data integration poses new computational challenges as well as exacerbates the ones associated with single-omics studies. Specialized computational approaches are required to effectively and efficiently perform integrative analysis of biomedical data acquired from diverse modalities. In this review, we discuss state-of-the-art ML-based approaches for tackling five specific computational challenges associated with integrative analysis: curse of dimensionality, data heterogeneity, missing data, class imbalance and scalability issues.

Computational Method for Identifying Malonylation Sites by Using Random Forest Algorithm

BACKGROUND

As a newly uncovered post-translational modification on the ε-amino group of lysine residue, protein malonylation was found to be involved in metabolic pathways and certain diseases. Apart from experimental approaches, several computational methods based on machine learning algorithms were recently proposed to predict malonylation sites. However, previous methods failed to address imbalanced data sizes between positive and negative samples.

OBJECTIVE

In this study, we identified the significant features of malonylation sites in a novel computational method which applied machine learning algorithms and balanced data sizes by applying synthetic minority over-sampling technique.

METHOD

Four types of features, namely, amino acid (AA) composition, position-specific scoring matrix (PSSM), AA factor, and disorder were used to encode residues in protein segments. Then, a two-step feature selection procedure including maximum relevance minimum redundancy and incremental feature selection, together with random forest algorithm, was performed on the constructed hybrid feature vector.

RESULTS

An optimal classifier was built from the optimal feature subset, which featured an F1-measure of 0.356. Feature analysis was performed on several selected important features.

CONCLUSION

Results showed that certain types of PSSM and disorder features may be closely associated with malonylation of lysine residues. Our study contributes to the development of computational approaches for predicting malonyllysine and provides insights into molecular mechanism of malonylation.

Classification of Widely and Rarely Expressed Genes with Recurrent Neural Network

A tissue-specific gene expression shapes the formation of tissues, while gene expression changes reflect the immune response of the human body to environmental stimulations or pressure, particularly in disease conditions, such as cancers. A few genes are commonly expressed across tissues or various cancers, while others are not. To investigate the functional differences between widely and rarely expressed genes, we defined the genes that were expressed in 32 normal tissues/cancers (i.e., called widely expressed genes; FPKM >1 in all samples) and those that were not detected (i.e., called rarely expressed genes; FPKM <1 in all samples) based on the large gene expression data set provided by Uhlen et al. Each gene was encoded using the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment scores. Minimum redundancy maximum relevance (mRMR) was used to measure and rank these features on the mRMR feature list. Thereafter, we applied the incremental feature selection method with a supervised classifier recurrent neural network (RNN) to select the discriminate features for classifying widely expressed genes from rarely expressed genes and construct an optimum RNN classifier. The Youden's indexes generated by the optimum RNN classifier and evaluated using a 10-fold cross validation were 0.739 for normal tissues and 0.639 for cancers. Furthermore, the underlying mechanisms of the key discriminate GO and KEGG features were analyzed. Results can facilitate the identification of the expression landscape of genes and elucidation of how gene expression shapes tissues and the microenvironment of cancers.

•

Some genes are widely expressed across tissues or various cancers.

•

A number of genes are rarely expressed across tissues or various cancers.

•

The functional differences between widely and rarely expressed genes were studied.

•

Several GO terms and KEGG pathways were extracted and analyzed.

HIV infection alters the human epigenetic landscape

Many complex diseases or traits are the results of both genetic and environmental factors. The environmental factors affect the human body by modifying its epigenetics, which controls the activity of genomes without mutating it. Viral infection is one of the common environmental factors for complex diseases. For example, the human immunodeficiency virus (HIV) infection can cause acquired immune deficiency syndrome (AIDS), HBV, and HCV infections are associated with hepatocellular carcinoma, and human papillomavirus infection is a causal factor in cervical carcinoma. In this study, to investigate how HIV infection affects DNA methylation, we analyzed the blood DNA methylation data of 485 512 sites in 44 HIV- and 142 HIV + patients. Several advanced computational methods were applied to identify the core distinctive features that were different between the HIV patients and the healthy controls. These methods can be used for differentiating HIV-infected patients from uninfected ones. These core distinctive DNA methylation features were confirmed to be functionally connected to premature aging and abnormal immune regulation, two typical pathological symptoms of HIV infection, revealing the potential regulatory mechanisms of HIV infection on the DNA methylation status of the host cells and provided novel insights on the pathogenesis of HIV infection and AIDS.

Tissue Expression Difference between mRNAs and lncRNAs

Messenger RNA (mRNA) and long noncoding RNA (lncRNA) are two main subgroups of RNAs participating in transcription regulation. With the development of next generation sequencing, increasing lncRNAs are identified. Many hidden functions of lncRNAs are also revealed. However, the differences in lncRNAs and mRNAs are still unclear. For example, we need to determine whether lncRNAs have stronger tissue specificity than mRNAs and which tissues have more lncRNAs expressed. To investigate such tissue expression difference between mRNAs and lncRNAs, we encoded 9339 lncRNAs and 14,294 mRNAs with 71 expression features, including 69 maximum expression features for 69 types of cells, one feature for the maximum expression in all cells, and one expression specificity feature that was measured as Chao-Shen-corrected Shannon's entropy. With advanced feature selection methods, such as maximum relevance minimum redundancy, incremental feature selection methods, and random forest algorithm, 13 features presented the dissimilarity of lncRNAs and mRNAs. The 11 cell subtype features indicated which cell types of the lncRNAs and mRNAs had the largest expression difference. Such cell subtypes may be the potential cell models for lncRNA identification and function investigation. The expression specificity feature suggested that the cell types to express mRNAs and lncRNAs were different. The maximum expression feature suggested that the maximum expression levels of mRNAs and lncRNAs were different. In addition, the rule learning algorithm, repeated incremental pruning to produce error reduction algorithm, was also employed to produce effective classification rules for classifying lncRNAs and mRNAs, which gave competitive results compared with random forest and could give a clearer picture of different expression patterns between lncRNAs and mRNAs. Results not only revealed the heterogeneous expression pattern of lncRNA and mRNA, but also gave rise to the development of a new tool to identify the potential biological functions of such RNA subgroups.

Identification of the copy number variant biomarkers for breast cancer subtypes

Breast cancer is a common and threatening malignant disease with multiple biological and clinical subtypes. It can be categorized into subtypes of luminal A, luminal B, Her2 positive, and basal-like. Copy number variants (CNVs) have been reported to be a potential and even better biomarker for cancer diagnosis than mRNA biomarkers, because it is considerably more stable and robust than gene expression. Thus, it is meaningful to detect CNVs of different cancers. To identify the CNV biomarker for breast cancer subtypes, we integrated the CNV data of more than 2000 samples from two large breast cancer databases, METABRIC and The Cancer Genome Atlas (TCGA). A Monte Carlo feature selection-based and incremental feature selection-based computational method was proposed and tested to identify the distinctive core CNVs in different breast cancer subtypes. We identified the CNV genes that may contribute to breast cancer tumorigenesis as well as built a set of quantitative distinctive rules for recognition of the breast cancer subtypes. The tenfold cross-validation Matthew's correlation coefficient (MCC) on METABRIC training set and the independent test on TCGA dataset were 0.515 and 0.492, respectively. The CNVs of PGAP3, GRB7, MIR4728, PNMT, STARD3, TCAP and ERBB2 were important for the accurate diagnosis of breast cancer subtypes. The findings reported in this study may further uncover the difference between different breast cancer subtypes and improve the diagnosis accuracy.

A Computational Method for Classifying Different Human Tissues with Quantitatively Tissue-Specific Expressed Genes

Tissue-specific gene expression has long been recognized as a crucial key for understanding tissue development and function. Efforts have been made in the past decade to identify tissue-specific expression profiles, such as the Human Proteome Atlas and FANTOM5. However, these studies mainly focused on "qualitatively tissue-specific expressed genes" which are highly enriched in one or a group of tissues but paid less attention to "quantitatively tissue-specific expressed genes", which are expressed in all or most tissues but with differential expression levels. In this study, we applied machine learning algorithms to build a computational method for identifying "quantitatively tissue-specific expressed genes" capable of distinguishing 25 human tissues from their expression patterns. Our results uncovered the expression of 432 genes as optimal features for tissue classification, which were obtained with a Matthews Correlation Coefficient (MCC) of more than 0.99 yielded by a support vector machine (SVM). This constructed model was superior to the SVM model using tissue enriched genes and yielded MCC of 0.985 on an independent test dataset, indicating its good generalization ability. These 432 genes were proven to be widely expressed in multiple tissues and a literature review of the top 23 genes found that most of them support their discriminating powers. As a complement to previous studies, our discovery of these quantitatively tissue-specific genes provides insights into the detailed understanding of tissue development and function.

An Efficient Feature Selection Strategy Based on Multiple Support Vector Machine Technology with Gene Expression Data

The application of gene expression data to the diagnosis and classification of cancer has become a hot issue in the field of cancer classification. Gene expression data usually contains a large number of tumor-free data and has the characteristics of high dimensions. In order to select determinant genes related to breast cancer from the initial gene expression data, we propose a new feature selection method, namely, support vector machine based on recursive feature elimination and parameter optimization (SVM-RFE-PO). The grid search (GS) algorithm, the particle swarm optimization (PSO) algorithm, and the genetic algorithm (GA) are applied to search the optimal parameters in the feature selection process. Herein, the new feature selection method contains three kinds of algorithms: support vector machine based on recursive feature elimination and grid search (SVM-RFE-GS), support vector machine based on recursive feature elimination and particle swarm optimization (SVM-RFE-PSO), and support vector machine based on recursive feature elimination and genetic algorithm (SVM-RFE-GA). Then the selected optimal feature subsets are used to train the SVM classifier for cancer classification. We also use random forest feature selection (RFFS), random forest feature selection and grid search (RFFS-GS), and minimal redundancy maximal relevance (mRMR) algorithm as feature selection methods to compare the effects of the SVM-RFE-PO algorithm. The results showed that the feature subset obtained by feature selection using SVM-RFE-PSO algorithm results has a better prediction performance of Area Under Curve (AUC) in the testing data set. This algorithm not only is time-saving, but also is capable of extracting more representative and useful genes.

Data mining of the cancer-related lncRNAs GO terms and KEGG pathways by using mRMR method

LncRNAs plays an important role in the regulation of gene expression. Identification of cancer-related lncRNAs GO terms and KEGG pathways is great helpful for revealing cancer-related functional biological processes. Therefore, in this study, we proposed a computational method to identify novel cancer-related lncRNAs GO terms and KEGG pathways. By using existing lncRNA database and Max-relevance Min-redundancy (mRMR) method, GO terms and KEGG pathways were evaluated based on their importance on distinguishing cancer-related and non-cancer-related lncRNAs. Finally, GO terms and KEGG pathways with high importance were presented and analyzed. Our literature reviewing showed that the top 10 ranked GO terms and pathways were really related to interpretable tumorigenesis according to recent publications.

Computational Approach to Investigating Key GO Terms and KEGG Pathways Associated with CNV

Choroidal neovascularization (CNV) is a severe eye disease that leads to blindness, especially in the elderly population. Various endogenous and exogenous regulatory factors promote its pathogenesis. However, the detailed molecular biological mechanisms of CNV have not been fully revealed. In this study, by using advanced computational tools, a number of key gene ontology (GO) terms and KEGG pathways were selected for CNV. A total of 29 validated genes associated with CNV and 17,639 nonvalidated genes were encoded based on the features derived from the GO terms and KEGG pathways by using the enrichment theory. The widely accepted feature selection method-maximum relevance and minimum redundancy (mRMR)-was applied to analyze and rank the features. An extensive literature review for the top 45 ranking features was conducted to confirm their close associations with CNV. Identifying the molecular biological mechanisms of CNV as described by the GO terms and KEGG pathways may contribute to improving the understanding of the pathogenesis of CNV.

Identification of Differentially Expressed Genes between Original Breast Cancer and Xenograft Using Machine Learning Algorithms

Breast cancer is one of the most common malignancies in women. Patient-derived tumor xenograft (PDX) model is a cutting-edge approach for drug research on breast cancer. However, PDX still exhibits differences from original human tumors, thereby challenging the molecular understanding of tumorigenesis. In particular, gene expression changes after tissues are transplanted from human to mouse model. In this study, we propose a novel computational method by incorporating several machine learning algorithms, including Monte Carlo feature selection (MCFS), random forest (RF), and rough set-based rule learning, to identify genes with significant expression differences between PDX and original human tumors. First, 831 breast tumors, including 657 PDX and 174 human tumors, were collected. Based on MCFS and RF, 32 genes were then identified to be informative for the prediction of PDX and human tumors and can be used to construct a prediction model. The prediction model exhibits a Matthews coefficient correlation value of 0.777. Seven interpretable interactions within the informative gene were detected based on the rough set-based rule learning. Furthermore, the seven interpretable interactions can be well supported by previous experimental studies. Our study not only presents a method for identifying informative genes with differential expression but also provides insights into the mechanism through which gene expression changes after being transplanted from human tumor into mouse model. This work would be helpful for research and drug development for breast cancer.

Identification and analysis of the cleavage site in a signal peptide using SMOTE, dagging, and feature selection methods

Several machine learning algorithms were adopted to investigate cleavage sites in a signal peptide. An optimal dagging based classifier was constructed and 870 important features were deemed to be important for this classifier.