A hybrid method for prediction and repositioning of drug Anatomical Therapeutic Chemical classes

PDATC-NCPMKL: Predicting drug's Anatomical Therapeutic Chemical (ATC) codes based on network consistency projection and multiple kernel learning

The development and discovery of new drugs is time-consuming and needs lots of human and material resources. Therefore, discovery of novel effects of existing drugs is an important alternative way, which can accelerate the process of designing "new" drugs. The anatomical Therapeutic Chemical (ATC) classification system recommended by World Health Organization (WHO) is a basic research area in this regard. A novel ATC code of an existing drug suggests its novel effects. Some computational models have been proposed, which can predict the drug-ATC code associations. However, their performance is not very high. There still exist spaces for improvement. In this study, a new recommendation system (named PDATC-NCPMKL), which incorporated network consistency projection and multi-kernel learning, was designed to identify drug-ATC code associations. For drugs or ATC codes, several kernels were constructed, which were fused by a multiple kernel learning method and an additional kernel integration scheme. To enhance the performance, the drug-ATC code association adjacency matrix was reformulated by a variant of weighted K nearest known neighbors (WKNKN). The reformulated adjacency matrix, drug and ATC code kernels were fed into network consistency projection to generate the association score matrix. The proposed recommendation system was tested on the ATC codes at the second, third and fourth levels in drug ATC classification system using ten-fold cross-validation. The results indicated that all AUROC and AUPR values were close to or exceeded 0.96. Such performance was higher than some existing computational models. Some additional tests were conducted to prove the utility of adjacency matrix reformulation and to analyze the importance of drug and ATC code kernels.

Drug-drug interaction prediction: databases, web servers and computational models

In clinical treatment, two or more drugs (i.e. drug combination) are simultaneously or successively used for therapy with the purpose of primarily enhancing the therapeutic efficacy or reducing drug side effects. However, inappropriate drug combination may not only fail to improve efficacy, but even lead to adverse reactions. Therefore, according to the basic principle of improving the efficacy and/or reducing adverse reactions, we should study drug-drug interactions (DDIs) comprehensively and thoroughly so as to reasonably use drug combination. In this review, we first introduced the basic conception and classification of DDIs. Further, some important publicly available databases and web servers about experimentally verified or predicted DDIs were briefly described. As an effective auxiliary tool, computational models for predicting DDIs can not only save the cost of biological experiments, but also provide relevant guidance for combination therapy to some extent. Therefore, we summarized three types of prediction models (including traditional machine learning-based models, deep learning-based models and score function-based models) proposed during recent years and discussed the advantages as well as limitations of them. Besides, we pointed out the problems that need to be solved in the future research of DDIs prediction and provided corresponding suggestions.

Identifying the kind behind SMILES-anatomical therapeutic chemical classification using structure-only representations

Anatomical Therapeutic Chemical (ATC) classification for compounds/drugs plays an important role in drug development and basic research. However, previous methods depend on interactions extracted from STITCH dataset which may make it depend on lab experiments. We present a pilot study to explore the possibility of conducting the ATC prediction solely based on the molecular structures. The motivation is to eliminate the reliance on the costly lab experiments so that the characteristics of a drug can be pre-assessed for better decision-making and effort-saving before the actual development. To this end, we construct a new benchmark consisting of 4545 compounds which is with larger scale than the one used in previous study. A light-weight prediction model is proposed. The model is with better explainability in the sense that it is consists of a straightforward tokenization that extracts and embeds statistically and physicochemically meaningful tokens, and a deep network backed by a set of pyramid kernels to capture multi-resolution chemical structural characteristics. Its efficacy has been validated in the experiments where it outperforms the state-of-the-art methods by 15.53% in accuracy and by 69.66% in terms of efficiency. We make the benchmark dataset, source code and web server open to ease the reproduction of this study.

Similarity-Based Method with Multiple-Feature Sampling for Predicting Drug Side Effects

Drugs can treat different diseases but also bring side effects. Undetected and unaccepted side effects for approved drugs can greatly harm the human body and bring huge risks for pharmaceutical companies. Traditional experimental methods used to determine the side effects have several drawbacks, such as low efficiency and high cost. One alternative to achieve this purpose is to design computational methods. Previous studies modeled a binary classification problem by pairing drugs and side effects; however, their classifiers can only extract one feature from each type of drug association. The present work proposed a novel multiple-feature sampling scheme that can extract several features from one type of drug association. Thirteen classification algorithms were employed to construct classifiers with features yielded by such scheme. Their performance was greatly improved compared with that of the classifiers that use the features yielded by the original scheme. Best performance was observed for the classifier based on random forest with MCC of 0.8661, AUROC of 0.969, and AUPR of 0.977. Finally, one key parameter in the multiple-feature sampling scheme was analyzed.

A convolutional neural network and graph convolutional network-based method for predicting the classification of anatomical therapeutic chemicals

MOTIVATION

The Anatomical Therapeutic Chemical (ATC) system is an official classification system established by the World Health Organization for medicines. Correctly assigning ATC classes to given compounds is an important research problem in drug discovery, which can not only discover the possible active ingredients of the compounds, but also infer theirs therapeutic, pharmacological and chemical properties.

RESULTS

In this article, we develop an end-to-end multi-label classifier called CGATCPred to predict 14 main ATC classes for given compounds. In order to extract rich features of each compound, we use the deep Convolutional Neural Network and shortcut connections to represent and learn the seven association scores between the given compound and others. Moreover, we construct the correlation graph of ATC classes and then apply graph convolutional network on the graph for label embedding abstraction. We use all label embedding to guide the learning process of compound representation. As a result, by using the Jackknife test, CGATCPred obtain reliable Aiming of 81.94%, Coverage of 82.88%, Accuracy 80.81%, Absolute True 76.58% and Absolute False 2.75%, yielding significantly improvements compared to exiting multi-label classifiers.

AVAILABILITY AND IMPLEMENTATION

The codes of CGATCPred are available at https://github.com/zhc940702/CGATCPred and https://zenodo.org/record/4552917.

Deep fusion learning facilitates anatomical therapeutic chemical recognition in drug repurposing and discovery

The advent of large-scale biomedical data and computational algorithms provides new opportunities for drug repurposing and discovery. It is of great interest to find an appropriate data representation and modeling method to facilitate these studies. The anatomical therapeutic chemical (ATC) classification system, proposed by the World Health Organization (WHO), is an essential source of information for drug repurposing and discovery. Besides, computational methods are applied to predict drug ATC classification. We conducted a systematic review of ATC computational prediction studies and revealed the differences in data sets, data representation, algorithm approaches, and evaluation metrics. We then proposed a deep fusion learning (DFL) framework to optimize the ATC prediction model, namely DeepATC. The methods based on graph convolutional network, inferring biological network and multimodel attentive fusion network were applied in DeepATC to extract the molecular topological information and low-dimensional representation from the molecular graph and heterogeneous biological networks. The results indicated that DeepATC achieved superior model performance with area under the curve (AUC) value at 0.968. Furthermore, the DFL framework was performed for the transcriptome data-based ATC prediction, as well as another independent task that is significantly relevant to drug discovery, namely drug-target interaction. The DFL-based model achieved excellent performance in the above-extended validation task, suggesting that the idea of aggregating the heterogeneous biological network and node's (molecule or protein) self-topological features will bring inspiration for broader drug repurposing and discovery research.

Analysis of the Impact of Medical Features and Risk Prediction of Acute Kidney Injury for Critical Patients Using Temporal Electronic Health Record Data With Attention-Based Neural Network

Acute kidney injury (AKI) is one of the most severe consequences of kidney injury, and it will also cause or aggravate the complications by the fast decline of kidney excretory function. Accurate AKI prediction, including the AKI case, AKI stage, and AKI onset time interval, can provide adequate support for effective interventions. Besides, discovering how the medical features affect the AKI result may also provide supporting information for disease treatment. An attention-based temporal neural network approach was employed in this study for AKI prediction and for the analysis of the impact of medical features from temporal electronic health record (EHR) data of patients before AKI diagnosis. We used the publicly available dataset provided by the Medical Information Mart for Intensive Care (MIMIC) for model training, validation, and testing, and then the model was applied in clinical practice. The improvement of AKI case prediction is around 5% AUC (area under the receiver operating characteristic curve), and the AUC value of AKI stage prediction on AKI stage 3 is over 82%. We also analyzed the data by two steps: the associations between the medical features and the AKI case (positive or inverse) and the extent of the impact of medical features on AKI prediction result. It shows that features, such as lactate, glucose, creatinine, blood urea nitrogen (BUN), prothrombin time (PT), and partial thromboplastin time (PTT), are positively associated with the AKI case, while there are inverse associations between the AKI case and features such as platelet, hemoglobin, hematocrit, urine, and international normalized ratio (INR). The laboratory test features such as urine, glucose, creatinine, sodium, and blood urea nitrogen and the medication features such as nonsteroidal anti-inflammatory drugs, agents acting on the renin–angiotensin system, and lipid-lowering medication were detected to have higher weights than other features in the proposed model, which may imply that these features have a great impact on the AKI case.

Drug repurposing against breast cancer by integrating drug-exposure expression profiles and drug-drug links based on graph neural network

MOTIVATION

Breast cancer is one of the leading causes of cancer deaths among women worldwide. It is necessary to develop new breast cancer drugs because of the shortcomings of existing therapies. The traditional discovery process is time-consuming and expensive. Repositioning of clinically approved drugs has emerged as a novel approach for breast cancer therapy. However, serendipitous or experiential repurposing cannot be used as a routine method.

RESULTS

In this study, we proposed a graph neural network model GraphRepur based on GraphSAGE for drug repurposing against breast cancer. GraphRepur integrated two major classes of computational methods, drug network-based and drug signature-based. The differentially expressed genes of disease, drug-exposure gene expression data and the drug-drug links information were collected. By extracting the drug signatures and topological structure information contained in the drug relationships, GraphRepur can predict new drugs for breast cancer, outperforming previous state-of-the-art approaches and some classic machine learning methods. The high-ranked drugs have indeed been reported as new uses for breast cancer treatment recently.

AVAILABILITYAND IMPLEMENTATION

The source code of our model and datasets are available at: https://github.com/cckamy/GraphRepur and https://figshare.com/articles/software/GraphRepur_Breast_Cancer_Drug_Repurposing/14220050.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

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.

Drug-drug similarity measure and its applications

Drug similarities play an important role in modern biology and medicine, as they help scientists gain deep insights into drugs' therapeutic mechanisms and conduct wet labs that may significantly improve the efficiency of drug research and development. Nowadays, a number of drug-related databases have been constructed, with which many methods have been developed for computing similarities between drugs for studying associations between drugs, human diseases, proteins (drug targets) and more. In this review, firstly, we briefly introduce the publicly available drug-related databases. Secondly, based on different drug features, interaction relationships and multimodal data, we summarize similarity calculation methods in details. Then, we discuss the applications of drug similarities in various biological and medical areas. Finally, we evaluate drug similarity calculation methods with common evaluation metrics to illustrate the important roles of drug similarity measures on different applications.

iATC-FRAKEL: a simple multi-label web server for recognizing anatomical therapeutic chemical classes of drugs with their fingerprints only

MOTIVATION

Anatomical therapeutic chemical (ATC) classification system is very important for drug utilization and studies. Correct prediction of the 14 classes in the first level for given drugs is an essential problem for the study on such system. Several multi-label classifiers have been proposed in this regard. However, only two of them provided the web servers and their performance was not very high. On the other hand, although some rest classifiers can provide better performance, they were built based on some prior knowledge on drugs, such as information of chemical-chemical interaction and chemical ontology, leading to limited applications. Furthermore, provided codes of these classifiers are almost inaccessible for pharmacologists.

RESULTS

In this study, we built a simple web server, namely iATC-FRAKEL. This web server only required the SMILES format of drugs as input and extracted their fingerprints for making prediction. The performance of the iATC-FRAKEL was much higher than all existing web servers and was comparable to the best multi-label classifier but had much wider applications. Such web server can be visited at http://cie.shmtu.edu.cn/iatc/index.

AVAILABILITY AND IMPLEMENTATION

The web server is available at http://cie.shmtu.edu.cn/iatc/index.

CONTACT

chen_lei1@163.com.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

iATC-NRAKEL: an efficient multi-label classifier for recognizing anatomical therapeutic chemical classes of drugs

MOTIVATION

The anatomical therapeutic chemical (ATC) classification system plays an increasingly important role in drug repositioning and discovery. The correct identification of classes in each level of such system that a given drug may belong to is an essential problem. Several multi-label classifiers have been proposed in this regard. Although they provided satisfactory performance, the feature extraction procedures were still rough. More refined features may further improve the predicted quality.

RESULTS

In this article, we provide a novel multi-label classifier, called iATC-NRAKEL, to predict drug ATC classes in the first level. To obtain more informative drug features, we employed the drug association information in STITCH and KEGG, which was organized by seven drug networks. The powerful network embedding algorithm, Mashup, was adopted to extract informative drug features. The obtained features were fed into the RAndom k-labELsets (RAKEL) algorithm with support vector machine as the basic classification algorithm to construct the classifier. The 10-fold cross-validation of the benchmark dataset with 3883 drugs showed that the accuracy and absolute true were 76.56 and 74.51%, respectively. The comparison results indicated that iATC-NRAKEL was much superior to all previous reported classifiers. Finally, the contribution of each network was analyzed.

AVAILABILITY AND IMPLEMENTATION

The codes of iATC-NRAKEL are available at https://github.com/zhou256/iATC-NRAKEL.

Relating Substructures and Side Effects of Drugs with Chemical-chemical Interactions

Background:

Drugs are very important for human life because they can provide treatment, cure, prevention, or diagnosis of different diseases. However, they also cause side effects, which can increase the risks for humans and pharmaceuticals companies. It is essential to identify drug side effects in drug discovery. To date, lots of computational methods have been proposed to predict the side effects of drugs and most of them used the fact that similar drugs always have similar side effects. However, previous studies did not analyze which substructures are highly related to which kind of side effect.

Method:

In this study, we conducted a computational investigation. In this regard, we extracted a drug set for each side effect, which consisted of drugs having the side effect. Also, for each substructure, a set was constructed by picking up drugs owing such substructure. The relationship between one side effect and one substructure was evaluated based on linkages between drugs in their corresponding drug sets, resulting in an Es value. Then, the statistical significance of Es value was measured by a permutation test.

Results and Conclusion:

A number of highly related pairs of side effects and substructures were obtained and some were extensively analyzed to confirm the reliability of the results reported in this study.

Drug Target Group Prediction with Multiple Drug Networks

Background:

Identification of drug-target interaction is essential in drug discovery. It is beneficial to predict unexpected therapeutic or adverse side effects of drugs. To date, several computational methods have been proposed to predict drug-target interactions because they are prompt and low-cost compared with traditional wet experiments.

Methods:

In this study, we investigated this problem in a different way. According to KEGG, drugs were classified into several groups based on their target proteins. A multi-label classification model was presented to assign drugs into correct target groups. To make full use of the known drug properties, five networks were constructed, each of which represented drug associations in one property. A powerful network embedding method, Mashup, was adopted to extract drug features from above-mentioned networks, based on which several machine learning algorithms, including RAndom k-labELsets (RAKEL) algorithm, Label Powerset (LP) algorithm and Support Vector Machine (SVM), were used to build the classification model.

Results and Conclusion:

Tenfold cross-validation yielded the accuracy of 0.839, exact match of 0.816 and hamming loss of 0.037, indicating good performance of the model. The contribution of each network was also analyzed. Furthermore, the network model with multiple networks was found to be superior to the one with a single network and classic model, indicating the superiority of the proposed model.

Prediction of Drug Side Effects with a Refined Negative Sample Selection Strategy

Drugs are an important way to treat various diseases. However, they inevitably produce side effects, bringing great risks to human bodies and pharmaceutical companies. How to predict the side effects of drugs has become one of the essential problems in drug research. Designing efficient computational methods is an alternative way. Some studies paired the drug and side effect as a sample, thereby modeling the problem as a binary classification problem. However, the selection of negative samples is a key problem in this case. In this study, a novel negative sample selection strategy was designed for accessing high-quality negative samples. Such strategy applied the random walk with restart (RWR) algorithm on a chemical-chemical interaction network to select pairs of drugs and side effects, such that drugs were less likely to have corresponding side effects, as negative samples. Through several tests with a fixed feature extraction scheme and different machine-learning algorithms, models with selected negative samples produced high performance. The best model even yielded nearly perfect performance. These models had much higher performance than those without such strategy or with another selection strategy. Furthermore, it is not necessary to consider the balance of positive and negative samples under such a strategy.

STS-NLSP: A Network-Based Label Space Partition Method for Predicting the Specificity of Membrane Transporter Substrates Using a Hybrid Feature of Structural and Semantic Similarity

Membrane transport proteins play crucial roles in the pharmacokinetics of substrate drugs, the drug resistance in cancer and are vital to the process of drug discovery, development and anti-cancer therapeutics. However, experimental methods to profile a substrate drug against a panel of transporters to determine its specificity are labor intensive and time consuming. In this article, we aim to develop an in silico multi-label classification approach to predict whether a substrate can specifically recognize one of the 13 categories of drug transporters ranging from ATP-binding cassette to solute carrier families using both structural fingerprints and chemical ontologies information of substrates. The data-driven network-based label space partition (NLSP) method was utilized to construct the model based on a hybrid of similarity-based feature by the integration of 2D fingerprint and semantic similarity. This method builds predictors for each label cluster (possibly intersecting) detected by community detection algorithms and takes union of label sets for a compound as final prediction. NLSP lies into the ensembles of multi-label classifier category in multi-label learning field. We utilized Cramér's V statistics to quantify the label correlations and depicted them via a heatmap. The jackknife tests and iterative stratification based cross-validation method were adopted on a benchmark dataset to evaluate the prediction performance of the proposed models both in multi-label and label-wise manner. Compared with other powerful multi-label methods, ML-kNN, MTSVM, and RAkELd, our multi-label classification model of NLPS-RF (random forest-based NLSP) has proven to be a feasible and effective model, and performed satisfactorily in the predictive task of transporter-substrate specificity. The idea behind NLSP method is intriguing and the power of NLSP remains to be explored for the multi-label learning problems in bioinformatics. The benchmark dataset, intermediate results and python code which can fully reproduce our experiments and results are available at https://github.com/dqwei-lab/STS.

Gene selection of rat hepatocyte proliferation using adaptive sparse group lasso with weighted gene co-expression network analysis

Grouped gene selection is the most important task for analyzing the microarray data of rat liver regeneration. Many existing gene selection methods cannot outstand the interactions among the selected genes. In the process of rat liver regeneration, one of the most important events involved in many biological processes is the proliferation of rat hepatocytes, so it can be used as a measure of the effectiveness of the method. Here we proposed an adaptive sparse group lasso to select genes in groups for rat hepatocyte proliferation. The weighted gene co-expression networks analysis was used to identify modules corresponding to gene pathways, based on which a strategy of dividing genes into groups was proposed. A strategy of adaptive gene selection was also presented by assessing the gene significance and introducing the adaptive lasso penalty. Moreover, an improved blockwise descent algorithm was proposed. Experimental results demonstrated that the proposed method can improve the classification accuracy, and select less number of significant genes which act jointly in groups and have direct or indirect effects on rat hepatocyte proliferation. The effectiveness of the method was verified by the method of rat hepatocyte proliferation.

A similarity-based method for prediction of drug side effects with heterogeneous information

Drugs can produce intended therapeutic effects to treat different diseases. However, they may also cause side effects at the same time. For an approved drug, it is best to detect all side effects it can produce. Otherwise, it may bring great risks for pharmaceuticals companies as well as be harmful to human body. It is urgent to design quick and reliable identification methods to detect the side effects for a given drug. In this study, a binary classification model was proposed to predict drug side effects. Different from most previous methods, our model termed the pair of drug and side effect as a sample and convert the original problem to a binary classification problem. Based on the similarity idea, each pair was represented by five features, each of which was derived from a type of drug property. The strong machine learning algorithm, random forest, was adopted as the prediction engine. The ten-fold cross-validation on five datasets with different negative samples indicated that the proposed model yielded a good performance of Matthews correlation coefficient around 0.550 and AUC around 0.8492. In addition, we also analyzed the contribution of each drug property for construction of the model. The results indicated that drug similarity in fingerprint was most related to the prediction of drug side effects and all drug properties gave less or more contributions.

A two-tiered unsupervised clustering approach for drug repositioning through heterogeneous data integration

Background

Drug repositioning is the process of identifying new uses for existing drugs. Computational drug repositioning methods can reduce the time, costs and risks of drug development by automating the analysis of the relationships in pharmacology networks. Pharmacology networks are large and heterogeneous. Clustering drugs into small groups can simplify large pharmacology networks, these subgroups can also be used as a starting point for repositioning drugs. In this paper, we propose a two-tiered drug-centric unsupervised clustering approach for drug repositioning, integrating heterogeneous drug data profiles: drug-chemical, drug-disease, drug-gene, drug-protein and drug-side effect relationships.

Results

The proposed drug repositioning approach is threefold; (i) clustering drugs based on their homogeneous profiles using the Growing Self Organizing Map (GSOM); (ii) clustering drugs based on drug-drug relation matrices based on the previous step, considering three state-of-the-art graph clustering methods; and (iii) inferring drug repositioning candidates and assigning a confidence value for each identified candidate. In this paper, we compare our two-tiered clustering approach against two existing heterogeneous data integration approaches with reference to the Anatomical Therapeutic Chemical (ATC) classification, using GSOM. Our approach yields Normalized Mutual Information (NMI) and Standardized Mutual Information (SMI) of 0.66 and 36.11, respectively, while the two existing methods yield NMI of 0.60 and 0.64 and SMI of 22.26 and 33.59. Moreover, the two existing approaches failed to produce useful cluster separations when using graph clustering algorithms while our approach is able to identify useful clusters for drug repositioning. Furthermore, we provide clinical evidence for four predicted results (Chlorthalidone, Indomethacin, Metformin and Thioridazine) to support that our proposed approach can be reliably used to infer ATC code and drug repositioning.

Conclusion

The proposed two-tiered unsupervised clustering approach is suitable for drug clustering and enables heterogeneous data integration. It also enables identifying reliable repositioning drug candidates with reference to ATC therapeutic classification. The repositioning drug candidates identified consistently by multiple clustering algorithms and with high confidence have a higher possibility of being effective repositioning candidates.

Electronic supplementary material

The online version of this article (10.1186/s12859-018-2123-4) contains supplementary material, which is available to authorized users.

Identification of gene expression signatures across different types of neural stem cells with the Monte-Carlo feature selection method

Adult neural stem cells (NSCs) are a group of multi-potent, self-renewing progenitor cells that contribute to the generation of new neurons and oligodendrocytes. Three subtypes of NSCs can be isolated based on the stages of the NSC lineage, including quiescent neural stem cells (qNSCs), activated neural stem cells (aNSCs) and neural progenitor cells (NPCs). Although it is widely accepted that these three groups of NSCs play different roles in the development of the nervous system, their molecular signatures are poorly understood. In this study, we applied the Monte-Carlo Feature Selection (MCFS) method to identify the gene expression signatures, which can yield a Matthews correlation coefficient (MCC) value of 0.918 with a support vector machine evaluated by ten-fold cross-validation. In addition, some classification rules yielded by the MCFS program for distinguishing above three subtypes were reported. Our results not only demonstrate a high classification capacity and subtype-specific gene expression patterns but also quantitatively reflect the pattern of the gene expression levels across the NSC lineage, providing insight into deciphering the molecular basis of NSC differentiation.

Inferring anatomical therapeutic chemical (ATC) class of drugs using shortest path and random walk with restart algorithms

The anatomical therapeutic chemical (ATC) classification system is a widely accepted drug classification scheme. This system comprises five levels and includes several classes in each level. Drugs are classified into classes according to their therapeutic effects and characteristics. The first level includes 14 main classes. In this study, we proposed two network-based models to infer novel potential chemicals deemed to belong in the first level of ATC classification. To build these models, two large chemical networks were constructed using the chemical-chemical interaction information retrieved from the Search Tool for Interactions of Chemicals (STITCH). Two classic network algorithms, shortest path (SP) and random walk with restart (RWR) algorithms, were executed on the corresponding network to mine novel chemicals for each ATC class using the validated drugs in a class as seed nodes. Then, the obtained chemicals yielded by these two algorithms were further evaluated by a permutation test and an association test. The former can exclude chemicals produced by the structure of the network, i.e., false positive discoveries. By contrast, the latter identifies the most important chemicals that have strong associations with the ATC class. Comparisons indicated that the two models can provide quite dissimilar results, suggesting that the results yielded by one model can be essential supplements for those obtained by the other model. In addition, several representative inferred chemicals were analyzed to confirm the reliability of the results generated by the two models. This article is part of a Special Issue entitled: Accelerating Precision Medicine through Genetic and Genomic Big Data Analysis edited by Yudong Cai & Tao Huang.

Identifying and analyzing different cancer subtypes using RNA-seq data of blood platelets

Detection and diagnosis of cancer are especially important for early prevention and effective treatments. Traditional methods of cancer detection are usually time-consuming and expensive. Liquid biopsy, a newly proposed noninvasive detection approach, can promote the accuracy and decrease the cost of detection according to a personalized expression profile. However, few studies have been performed to analyze this type of data, which can promote more effective methods for detection of different cancer subtypes. In this study, we applied some reliable machine learning algorithms to analyze data retrieved from patients who had one of six cancer subtypes (breast cancer, colorectal cancer, glioblastoma, hepatobiliary cancer, lung cancer and pancreatic cancer) as well as healthy persons. Quantitative gene expression profiles were used to encode each sample. Then, they were analyzed by the maximum relevance minimum redundancy method. Two feature lists were obtained in which genes were ranked rigorously. The incremental feature selection method was applied to the mRMR feature list to extract the optimal feature subset, which can be used in the support vector machine algorithm to determine the best performance for the detection of cancer subtypes and healthy controls. The ten-fold cross-validation for the constructed optimal classification model yielded an overall accuracy of 0.751. On the other hand, we extracted the top eighteen features (genes), including TTN, RHOH, RPS20, TRBC2, in another feature list, the MaxRel feature list, and performed a detailed analysis of them. The results indicated that these genes could be important biomarkers for discriminating different cancer subtypes and healthy controls.

A computational method for the identification of candidate drugs for non-small cell lung cancer

Lung cancer causes a large number of deaths per year. Until now, a cure for this disease has not been found or developed. Finding an effective drug through traditional experimental methods invariably costs millions of dollars and takes several years. It is imperative that computational methods be developed to integrate several types of existing information to identify candidate drugs for further study, which could reduce the cost and time of development. In this study, we tried to advance this effort by proposing a computational method to identify candidate drugs for non-small cell lung cancer (NSCLC), a major type of lung cancer. The method used three steps: (1) preliminary screening, (2) screening compounds by an association test and a permutation test, (3) screening compounds using an EM clustering algorithm. In the first step, based on the chemical-chemical interaction information reported in STITCH, a well-known database that reports interactions between chemicals and proteins, and approved NSCLC drugs, compounds that can interact with at least one approved NSCLC drug were picked. In the second step, the association test selected compounds that can interact with at least one NSCLC-related chemical and at least one NSCLC-related gene, and subsequently, the permutation test was used to discard nonspecific compounds from the remaining compounds. In the final step, core compounds were selected using a powerful clustering algorithm, the EM algorithm. Six putative compounds, protoporphyrin IX, hematoporphyrin, canertinib, lapatinib, pelitinib, and dacomitinib, were identified by this method. Previously published data show that all of the selected compounds have been reported to possess anti-NSCLC activity, indicating high probabilities of these compounds being novel candidate drugs for NSCLC.

Predicting anatomic therapeutic chemical classification codes using tiered learning

BACKGROUND

The low success rate and high cost of drug discovery requires the development of new paradigms to identify molecules of therapeutic value. The Anatomical Therapeutic Chemical (ATC) Code System is a World Health Organization (WHO) proposed classification that assigns multi-level codes to compounds based on their therapeutic, pharmacological and chemical characteristics as well as the in-vivo sites(s) of activity. The ability to predict ATC codes of compounds can assist in creation of high-quality chemical libraries for drug screening and in applications such as drug repositioning. We propose a machine learning architecture called tiered learning for prediction of ATC codes that relies on the prediction results of the higher levels of the ATC code to simplify the predictions of the lower levels.

RESULTS

The proposed approach was validated using a number of compounds in both cross-validation and test setting. The validation experiments compared chemical descriptors, initialization methods and classification algorithms. The prediction accuracy obtained with tiered learning was found to be either comparable or better than that of established methods. Additionally, the experiments demonstrated the generalizability of the tiered learning architecture, in that its use was found to improve prediction rates for a majority of machine learning algorithms when compared to their stand-alone application.

CONCLUSION

The basis of our approach lies in the observation that anatomical-therapeutic biological activity of certain types typically precludes activities of many other types. Thus, there exists a characteristic distribution of the ATC codes, which can be leveraged to limit the search-space of possible codes that can be ascribed at a particular level once the codes at the preceding levels are known. Tiered learning utilizes this observation to constrain the learning space for ATC codes at a particular level based on the ATC code at higher levels. This simplifies the prediction and allows for improved accuracy.

iATC-mHyb: a hybrid multi-label classifier for predicting the classification of anatomical therapeutic chemicals

Recommended by the World Health Organization (WHO), drug compounds have been classified into 14 main ATC (Anatomical Therapeutic Chemical) classes according to their therapeutic and chemical characteristics. Given an uncharacterized compound, can we develop a computational method to fast identify which ATC class or classes it belongs to? The information thus obtained will timely help adjusting our focus and selection, significantly speeding up the drug development process. But this problem is by no means an easy one since some drug compounds may belong to two or more than two ATC classes. To address this problem, using the DO (Drug Ontology) approach based on the ChEBI (Chemical Entities of Biological Interest) database, we developed a predictor called iATC-mDO. Subsequently, hybridizing it with an existing drug ATC classifier, we constructed a predictor called iATC-mHyb. It has been demonstrated by the rigorous cross-validation and from five different measuring angles that iATC-mHyb is remarkably superior to the best existing predictor in identifying the ATC classes for drug compounds. To convenience most experimental scientists, a user-friendly web-server for iATC-mHyd has been established at http://www.jci-bioinfo.cn/iATC-mHyb, by which users can easily get their desired results without the need to go through the complicated mathematical equations involved.

Identification of Candidate Genes Related to Inflammatory Bowel Disease Using Minimum Redundancy Maximum Relevance, Incremental Feature Selection, and the Shortest-Path Approach

Identification of disease genes is a hot topic in biomedicine and genomics. However, it is a challenging problem because of the complexity of diseases. Inflammatory bowel disease (IBD) is an idiopathic disease caused by a dysregulated immune response to host intestinal microflora. It has been proven to be associated with the development of intestinal malignancies. Although the specific pathological characteristics and genetic background of IBD have been partially revealed, it is still an overdetermined disease and the blueprint of all genetic variants still needs to be improved. In this study, a novel computational method was built to identify genes related to IBD. Samples from two subtypes of IBD (ulcerative colitis and Crohn's disease) and normal samples were employed. By analyzing the gene expression profiles of these samples using minimum redundancy maximum relevance and incremental feature selection, 21 genes were obtained that could effectively distinguish samples from the two subtypes of IBD and the normal samples. Then, the shortest-path approach was used to search for an additional 20 genes in a large network constructed using protein-protein interactions based on the above-mentioned 21 genes. Analyses of the 41 genes obtained indicate that they are closely associated with this disease.

Exploring Mouse Protein Function via Multiple Approaches

Although the number of available protein sequences is growing exponentially, functional protein annotations lag far behind. Therefore, accurate identification of protein functions remains one of the major challenges in molecular biology. In this study, we presented a novel approach to predict mouse protein functions. The approach was a sequential combination of a similarity-based approach, an interaction-based approach and a pseudo amino acid composition-based approach. The method achieved an accuracy of about 0.8450 for the 1st-order predictions in the leave-one-out and ten-fold cross-validations. For the results yielded by the leave-one-out cross-validation, although the similarity-based approach alone achieved an accuracy of 0.8756, it was unable to predict the functions of proteins with no homologues. Comparatively, the pseudo amino acid composition-based approach alone reached an accuracy of 0.6786. Although the accuracy was lower than that of the previous approach, it could predict the functions of almost all proteins, even proteins with no homologues. Therefore, the combined method balanced the advantages and disadvantages of both approaches to achieve efficient performance. Furthermore, the results yielded by the ten-fold cross-validation indicate that the combined method is still effective and stable when there are no close homologs are available. However, the accuracy of the predicted functions can only be determined according to known protein functions based on current knowledge. Many protein functions remain unknown. By exploring the functions of proteins for which the 1st-order predicted functions are wrong but the 2nd-order predicted functions are correct, the 1st-order wrongly predicted functions were shown to be closely associated with the genes encoding the proteins. The so-called wrongly predicted functions could also potentially be correct upon future experimental verification. Therefore, the accuracy of the presented method may be much higher in reality.

The Use of Gene Ontology Term and KEGG Pathway Enrichment for Analysis of Drug Half-Life

A drug's biological half-life is defined as the time required for the human body to metabolize or eliminate 50% of the initial drug dosage. Correctly measuring the half-life of a given drug is helpful for the safe and accurate usage of the drug. In this study, we investigated which gene ontology (GO) terms and biological pathways were highly related to the determination of drug half-life. The investigated drugs, with known half-lives, were analyzed based on their enrichment scores for associated GO terms and KEGG pathways. These scores indicate which GO terms or KEGG pathways the drug targets. The feature selection method, minimum redundancy maximum relevance, was used to analyze these GO terms and KEGG pathways and to identify important GO terms and pathways, such as sodium-independent organic anion transmembrane transporter activity (GO:0015347), monoamine transmembrane transporter activity (GO:0008504), negative regulation of synaptic transmission (GO:0050805), neuroactive ligand-receptor interaction (hsa04080), serotonergic synapse (hsa04726), and linoleic acid metabolism (hsa00591), among others. This analysis confirmed our results and may show evidence for a new method in studying drug half-lives and building effective computational methods for the prediction of drug half-lives.

Identification of compound-protein interactions through the analysis of gene ontology, KEGG enrichment for proteins and molecular fragments of compounds

Compound-protein interactions play important roles in every cell via the recognition and regulation of specific functional proteins. The correct identification of compound-protein interactions can lead to a good comprehension of this complicated system and provide useful input for the investigation of various attributes of compounds and proteins. In this study, we attempted to understand this system by extracting properties from both proteins and compounds, in which proteins were represented by gene ontology and KEGG pathway enrichment scores and compounds were represented by molecular fragments. Advanced feature selection methods, including minimum redundancy maximum relevance, incremental feature selection, and the basic machine learning algorithm random forest, were used to analyze these properties and extract core factors for the determination of actual compound-protein interactions. Compound-protein interactions reported in The Binding Databases were used as positive samples. To improve the reliability of the results, the analytic procedure was executed five times using different negative samples. Simultaneously, five optimal prediction methods based on a random forest and yielding maximum MCCs of approximately 77.55 % were constructed and may be useful tools for the prediction of compound-protein interactions. This work provides new clues to understanding the system of compound-protein interactions by analyzing extracted core features. Our results indicate that compound-protein interactions are related to biological processes involving immune, developmental and hormone-associated pathways.

Analysis and prediction of drug-drug interaction by minimum redundancy maximum relevance and incremental feature selection

Drug-drug interaction (DDI) defines a situation in which one drug affects the activity of another when both are administered together. DDI is a common cause of adverse drug reactions and sometimes also leads to improved therapeutic effects. Therefore, it is of great interest to discover novel DDIs according to their molecular properties and mechanisms in a robust and rigorous way. This paper attempts to predict effective DDIs using the following properties: (1) chemical interaction between drugs; (2) protein interactions between the targets of drugs; and (3) target enrichment of KEGG pathways. The data consisted of 7323 pairs of DDIs collected from the DrugBank and 36,615 pairs of drugs constructed by randomly combining two drugs. Each drug pair was represented by 465 features derived from the aforementioned three categories of properties. The random forest algorithm was adopted to train the prediction model. Some feature selection techniques, including minimum redundancy maximum relevance and incremental feature selection, were used to extract key features as the optimal input for the prediction model. The extracted key features may help to gain insights into the mechanisms of DDIs and provide some guidelines for the relevant clinical medication developments, and the prediction model can give new clues for identification of novel DDIs.

Identification of new candidate drugs for lung cancer using chemical-chemical interactions, chemical-protein interactions and a K-means clustering algorithm

Lung cancer, characterized by uncontrolled cell growth in the lung tissue, is the leading cause of global cancer deaths. Until now, effective treatment of this disease is limited. Many synthetic compounds have emerged with the advancement of combinatorial chemistry. Identification of effective lung cancer candidate drug compounds among them is a great challenge. Thus, it is necessary to build effective computational methods that can assist us in selecting for potential lung cancer drug compounds. In this study, a computational method was proposed to tackle this problem. The chemical-chemical interactions and chemical-protein interactions were utilized to select candidate drug compounds that have close associations with approved lung cancer drugs and lung cancer-related genes. A permutation test and K-means clustering algorithm were employed to exclude candidate drugs with low possibilities to treat lung cancer. The final analysis suggests that the remaining drug compounds have potential anti-lung cancer activities and most of them have structural dissimilarity with approved drugs for lung cancer.

Analysis and Identification of Aptamer-Compound Interactions with a Maximum Relevance Minimum Redundancy and Nearest Neighbor Algorithm

The development of biochemistry and molecular biology has revealed an increasingly important role of compounds in several biological processes. Like the aptamer-protein interaction, aptamer-compound interaction attracts increasing attention. However, it is time-consuming to select proper aptamers against compounds using traditional methods, such as exponential enrichment. Thus, there is an urgent need to design effective computational methods for searching effective aptamers against compounds. This study attempted to extract important features for aptamer-compound interactions using feature selection methods, such as Maximum Relevance Minimum Redundancy, as well as incremental feature selection. Each aptamer-compound pair was represented by properties derived from the aptamer and compound, including frequencies of single nucleotides and dinucleotides for the aptamer, as well as the constitutional, electrostatic, quantum-chemical, and space conformational descriptors of the compounds. As a result, some important features were obtained. To confirm the importance of the obtained features, we further discussed the associations between them and aptamer-compound interactions. Simultaneously, an optimal prediction model based on the nearest neighbor algorithm was built to identify aptamer-compound interactions, which has the potential to be a useful tool for the identification of novel aptamer-compound interactions. The program is available upon the request.

The use of Gene Ontology terms and KEGG pathways for analysis and prediction of oncogenes

BACKGROUND

Oncogenes are a type of genes that have the potential to cause cancer. Most normal cells undergo programmed cell death, namely apoptosis, but activated oncogenes can help cells avoid apoptosis and survive. Thus, studying oncogenes is helpful for obtaining a good understanding of the formation and development of various types of cancers.

METHODS

In this study, we proposed a computational method, called OPM, for investigating oncogenes from the view of Gene Ontology (GO) and biological pathways. All investigated genes, including validated oncogenes retrieved from some public databases and other genes that have not been reported to be oncogenes thus far, were encoded into numeric vectors according to the enrichment theory of GO terms and KEGG pathways. Some popular feature selection methods, minimum redundancy maximum relevance and incremental feature selection, and an advanced machine learning algorithm, random forest, were adopted to analyze the numeric vectors to extract key GO terms and KEGG pathways.

RESULTS

Along with the oncogenes, GO terms and KEGG pathways were discussed in terms of their relevance in this study. Some important GO terms and KEGG pathways were extracted using feature selection methods and were confirmed to be highly related to oncogenes. Additionally, the importance of these terms and pathways in predicting oncogenes was further demonstrated by finding new putative oncogenes based on them.

CONCLUSIONS

This study investigated oncogenes based on GO terms and KEGG pathways. Some important GO terms and KEGG pathways were confirmed to be highly related to oncogenes. We hope that these GO terms and KEGG pathways can provide new insight for the study of oncogenes, particularly for building more effective prediction models to identify novel oncogenes. The program is available upon request.

GENERAL SIGNIFICANCE

We hope that the new findings listed in this study may provide a new insight for the investigation of oncogenes. This article is part of a Special Issue entitled "System Genetics" Guest Editor: Dr. Yudong Cai and Dr. Tao Huang.

Mining for novel tumor suppressor genes using a shortest path approach

Cancer, being among the most serious diseases, causes many deaths every year. Many investigators have devoted themselves to designing effective treatments for this disease. Cancer always involves abnormal cell growth with the potential to invade or spread to other parts of the body. In contrast, tumor suppressor genes (TSGs) act as guardians to prevent a disordered cell cycle and genomic instability in normal cells. Studies on TSGs can assist in the design of effective treatments against cancer. In this study, we propose a computational method to discover potential TSGs. Based on the known TSGs, a number of candidate genes were selected by applying the shortest path approach in a weighted graph that was constructed using protein-protein interaction network. The analysis of selected genes shows that some of them are new TSGs recently reported in the literature, while others may be novel TSGs.

Identifying New Candidate Genes and Chemicals Related to Prostate Cancer Using a Hybrid Network and Shortest Path Approach

Prostate cancer is a type of cancer that occurs in the male prostate, a gland in the male reproductive system. Because prostate cancer cells may spread to other parts of the body and can influence human reproduction, understanding the mechanisms underlying this disease is critical for designing effective treatments. The identification of as many genes and chemicals related to prostate cancer as possible will enhance our understanding of this disease. In this study, we proposed a computational method to identify new candidate genes and chemicals based on currently known genes and chemicals related to prostate cancer by applying a shortest path approach in a hybrid network. The hybrid network was constructed according to information concerning chemical-chemical interactions, chemical-protein interactions, and protein-protein interactions. Many of the obtained genes and chemicals are associated with prostate cancer.

A computational method for the identification of new candidate carcinogenic and non-carcinogenic chemicals

Cancer is one of the leading causes of human death. Based on current knowledge, one of the causes of cancer is exposure to toxic chemical compounds, including radioactive compounds, dioxin, and arsenic. The identification of new carcinogenic chemicals may warn us of potential danger and help to identify new ways to prevent cancer. In this study, a computational method was proposed to identify potential carcinogenic chemicals, as well as non-carcinogenic chemicals. According to the current validated carcinogenic and non-carcinogenic chemicals from the CPDB (Carcinogenic Potency Database), the candidate chemicals were searched in a weighted chemical network constructed according to chemical-chemical interactions. Then, the obtained candidate chemicals were further selected by a randomization test and information on chemical interactions and structures. The analyses identified several candidate carcinogenic chemicals, while those candidates identified as non-carcinogenic were supported by a literature search. In addition, several candidate carcinogenic/non-carcinogenic chemicals exhibit structural dissimilarity with validated carcinogenic/non-carcinogenic chemicals.

Identification of New Candidate Genes and Chemicals Related to Esophageal Cancer Using a Hybrid Interaction Network of Chemicals and Proteins

Cancer is a serious disease responsible for many deaths every year in both developed and developing countries. One reason is that the mechanisms underlying most types of cancer are still mysterious, creating a great block for the design of effective treatments. In this study, we attempted to clarify the mechanism underlying esophageal cancer by searching for novel genes and chemicals. To this end, we constructed a hybrid network containing both proteins and chemicals, and generalized an existing computational method previously used to identify disease genes to identify new candidate genes and chemicals simultaneously. Based on jackknife test, our generalized method outperforms or at least performs at the same level as those obtained by a widely used method - the Random Walk with Restart (RWR). The analysis results of the final obtained genes and chemicals demonstrated that they highly shared gene ontology (GO) terms and KEGG pathways with direct and indirect associations with esophageal cancer. In addition, we also discussed the likelihood of selected candidate genes and chemicals being novel genes and chemicals related to esophageal cancer.

The Use of Chemical-Chemical Interaction and Chemical Structure to Identify New Candidate Chemicals Related to Lung Cancer

Lung cancer causes over one million deaths every year worldwide. However, prevention and treatment methods for this serious disease are limited. The identification of new chemicals related to lung cancer may aid in disease prevention and the design of more effective treatments. This study employed a weighted network, constructed using chemical-chemical interaction information, to identify new chemicals related to two types of lung cancer: non-small lung cancer and small-cell lung cancer. Then, a randomization test as well as chemical-chemical interaction and chemical structure information were utilized to make further selections. A final analysis of these new chemicals in the context of the current literature indicates that several chemicals are strongly linked to lung cancer.

Gene Ontology and KEGG Pathway Enrichment Analysis of a Drug Target-Based Classification System

Drug-target interaction (DTI) is a key aspect in pharmaceutical research. With the ever-increasing new drug data resources, computational approaches have emerged as powerful and labor-saving tools in predicting new DTIs. However, so far, most of these predictions have been based on structural similarities rather than biological relevance. In this study, we proposed for the first time a "GO and KEGG enrichment score" method to represent a certain category of drug molecules by further classification and interpretation of the DTI database. A benchmark dataset consisting of 2,015 drugs that are assigned to nine categories ((1) G protein-coupled receptors, (2) cytokine receptors, (3) nuclear receptors, (4) ion channels, (5) transporters, (6) enzymes, (7) protein kinases, (8) cellular antigens and (9) pathogens) was constructed by collecting data from KEGG. We analyzed each category and each drug for its contribution in GO terms and KEGG pathways using the popular feature selection "minimum redundancy maximum relevance (mRMR)" method, and key GO terms and KEGG pathways were extracted. Our analysis revealed the top enriched GO terms and KEGG pathways of each drug category, which were highly enriched in the literature and clinical trials. Our results provide for the first time the biological relevance among drugs, targets and biological functions, which serves as a new basis for future DTI predictions.

Identification of novel thyroid cancer-related genes and chemicals using shortest path algorithm

Thyroid cancer is a typical endocrine malignancy. In the past three decades, the continued growth of its incidence has made it urgent to design effective treatments to treat this disease. To this end, it is necessary to uncover the mechanism underlying this disease. Identification of thyroid cancer-related genes and chemicals is helpful to understand the mechanism of thyroid cancer. In this study, we generalized some previous methods to discover both disease genes and chemicals. The method was based on shortest path algorithm and applied to discover novel thyroid cancer-related genes and chemicals. The analysis of the final obtained genes and chemicals suggests that some of them are crucial to the formation and development of thyroid cancer. It is indicated that the proposed method is effective for the discovery of novel disease genes and chemicals.

Prediction of drug indications based on chemical interactions and chemical similarities

Discovering potential indications of novel or approved drugs is a key step in drug development. Previous computational approaches could be categorized into disease-centric and drug-centric based on the starting point of the issues or small-scaled application and large-scale application according to the diversity of the datasets. Here, a classifier has been constructed to predict the indications of a drug based on the assumption that interactive/associated drugs or drugs with similar structures are more likely to target the same diseases using a large drug indication dataset. To examine the classifier, it was conducted on a dataset with 1,573 drugs retrieved from Comprehensive Medicinal Chemistry database for five times, evaluated by 5-fold cross-validation, yielding five 1st order prediction accuracies that were all approximately 51.48%. Meanwhile, the model yielded an accuracy rate of 50.00% for the 1st order prediction by independent test on a dataset with 32 other drugs in which drug repositioning has been confirmed. Interestingly, some clinically repurposed drug indications that were not included in the datasets are successfully identified by our method. These results suggest that our method may become a useful tool to associate novel molecules with new indications or alternative indications with existing drugs.

Discovery of new candidate genes related to brain development using protein interaction information

Human brain development is a dramatic process composed of a series of complex and fine-tuned spatiotemporal gene expressions. A good comprehension of this process can assist us in developing the potential of our brain. However, we have only limited knowledge about the genes and gene functions that are involved in this biological process. Therefore, a substantial demand remains to discover new brain development-related genes and identify their biological functions. In this study, we aimed to discover new brain-development related genes by building a computational method. We referred to a series of computational methods used to discover new disease-related genes and developed a similar method. In this method, the shortest path algorithm was executed on a weighted graph that was constructed using protein-protein interactions. New candidate genes fell on at least one of the shortest paths connecting two known genes that are related to brain development. A randomization test was then adopted to filter positive discoveries. Of the final identified genes, several have been reported to be associated with brain development, indicating the effectiveness of the method, whereas several of the others may have potential roles in brain development.

Protein submitochondrial localization from integrated sequence representation and SVM-based backward feature extraction

Mitochondrion, a tiny energy factory, plays an important role in various biological processes of most eukaryotic cells.