1
|
Singh J, Khanna NN, Rout RK, Singh N, Laird JR, Singh IM, Kalra MK, Mantella LE, Johri AM, Isenovic ER, Fouda MM, Saba L, Fatemi M, Suri JS. GeneAI 3.0: powerful, novel, generalized hybrid and ensemble deep learning frameworks for miRNA species classification of stationary patterns from nucleotides. Sci Rep 2024; 14:7154. [PMID: 38531923 PMCID: PMC11344070 DOI: 10.1038/s41598-024-56786-9] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/11/2023] [Accepted: 03/11/2024] [Indexed: 03/28/2024] Open
Abstract
Due to the intricate relationship between the small non-coding ribonucleic acid (miRNA) sequences, the classification of miRNA species, namely Human, Gorilla, Rat, and Mouse is challenging. Previous methods are not robust and accurate. In this study, we present AtheroPoint's GeneAI 3.0, a powerful, novel, and generalized method for extracting features from the fixed patterns of purines and pyrimidines in each miRNA sequence in ensemble paradigms in machine learning (EML) and convolutional neural network (CNN)-based deep learning (EDL) frameworks. GeneAI 3.0 utilized five conventional (Entropy, Dissimilarity, Energy, Homogeneity, and Contrast), and three contemporary (Shannon entropy, Hurst exponent, Fractal dimension) features, to generate a composite feature set from given miRNA sequences which were then passed into our ML and DL classification framework. A set of 11 new classifiers was designed consisting of 5 EML and 6 EDL for binary/multiclass classification. It was benchmarked against 9 solo ML (SML), 6 solo DL (SDL), 12 hybrid DL (HDL) models, resulting in a total of 11 + 27 = 38 models were designed. Four hypotheses were formulated and validated using explainable AI (XAI) as well as reliability/statistical tests. The order of the mean performance using accuracy (ACC)/area-under-the-curve (AUC) of the 24 DL classifiers was: EDL > HDL > SDL. The mean performance of EDL models with CNN layers was superior to that without CNN layers by 0.73%/0.92%. Mean performance of EML models was superior to SML models with improvements of ACC/AUC by 6.24%/6.46%. EDL models performed significantly better than EML models, with a mean increase in ACC/AUC of 7.09%/6.96%. The GeneAI 3.0 tool produced expected XAI feature plots, and the statistical tests showed significant p-values. Ensemble models with composite features are highly effective and generalized models for effectively classifying miRNA sequences.
Collapse
Affiliation(s)
- Jaskaran Singh
- Department of Computer Science, Graphic Era Deemed to be University, Dehradun, Uttarakhand, India
| | - Narendra N Khanna
- Department of Cardiology, Indraprastha APOLLO Hospitals, New Delhi, India
| | - Ranjeet K Rout
- Department of Computer Science and Engineering, NIT Srinagar, Hazratbal, Srinagar, India
| | - Narpinder Singh
- Department of Food Science, Graphic Era Deemed to be University, Dehradun, Uttarakhand, India
| | - John R Laird
- Heart and Vascular Institute, Adventist Health St. Helena, St Helena, CA, USA
| | - Inder M Singh
- Advanced Cardiac and Vascular Institute, Sacramento, CA, USA
| | - Mannudeep K Kalra
- Department of Radiology, Massachusetts General Hospital, Boston, MA, 02115, USA
| | - Laura E Mantella
- Department of Biomedical and Molecular Sciences, Queen's University, Kingston, ON, Canada
| | - Amer M Johri
- Department of Biomedical and Molecular Sciences, Queen's University, Kingston, ON, Canada
| | - Esma R Isenovic
- Laboratory for Molecular Genetics and Radiobiology, University of Belgrade, Belgrade, Serbia
| | - Mostafa M Fouda
- Department of Electrical and Computer Engineering, Idaho State University, Pocatello, ID, 83209, USA
| | - Luca Saba
- Department of Neurology, University of Cagliari, Cagliari, Italy
| | - Mostafa Fatemi
- Department of Physiology and Biomedical Engineering, Mayo Clinic, Rochester, MN, 55905, USA
| | - Jasjit S Suri
- Stroke Monitoring and Diagnostic Division, AtheroPoint LLC, Roseville, CA, 95661, USA.
| |
Collapse
|
2
|
Abstract
Tiny single-stranded noncoding RNAs with size 19-27 nucleotides serve as microRNAs (miRNAs), which have emerged as key gene regulators in the last two decades. miRNAs serve as one of the hallmarks in regulatory pathways with critical roles in human diseases. Ever since the discovery of miRNAs, researchers have focused on how mature miRNAs are produced from precursor mRNAs. Experimental methods are faced with notorious challenges in terms of experimental design, since it is time consuming and not cost-effective. Hence, different computational methods have been employed for the identification of miRNA sequences where most of them labeled as miRNA predictors are in fact pre-miRNA predictors and provide no information about the putative miRNA location within the pre-miRNA. This chapter provides an update and the current state of the art in this area covering various methods and 15 software suites used for prediction of mature miRNA.
Collapse
Affiliation(s)
- Malik Yousef
- Department of Information System, Galilee Digital Health Research Center (GDH), Zefat Academic College, Zefat, Israel
| | - Alisha Parveen
- Rudolf‑Zenker Institute of Experimental Surgery, Rostock University Medical Center, Rostock, Germany
| | - Abhishek Kumar
- Institute of Bioinformatics, Bangalore, India. .,Manipal Academy of Higher Education (MAHE), Manipal, Karnataka, India.
| |
Collapse
|
3
|
Velandia-Huerto CA, Fallmann J, Stadler PF. miRNAture-Computational Detection of microRNA Candidates. Genes (Basel) 2021; 12:348. [PMID: 33673400 PMCID: PMC7996739 DOI: 10.3390/genes12030348] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/01/2021] [Revised: 02/19/2021] [Accepted: 02/20/2021] [Indexed: 12/16/2022] Open
Abstract
Homology-based annotation of short RNAs, including microRNAs, is a difficult problem because their inherently small size limits the available information. Highly sensitive methods, including parameter optimized blast, nhmmer, or cmsearch runs designed to increase sensitivity inevitable lead to large numbers of false positives, which can be detected only by detailed analysis of specific features typical for a RNA family and/or the analysis of conservation patterns in structure-annotated multiple sequence alignments. The miRNAture pipeline implements a workflow specific to animal microRNAs that automatizes homology search and validation steps. The miRNAture pipeline yields very good results for a large number of "typical" miRBase families. However, it also highlights difficulties with atypical cases, in particular microRNAs deriving from repetitive elements and microRNAs with unusual, branched precursor structures and atypical locations of the mature product, which require specific curation by domain experts.
Collapse
Affiliation(s)
- Cristian A. Velandia-Huerto
- Bioinformatics Group, Department of Computer Science, and Interdisciplinary Center for Bioinformatics, Leipzig University, D-04107 Leipzig, Germany
| | - Jörg Fallmann
- Bioinformatics Group, Department of Computer Science, and Interdisciplinary Center for Bioinformatics, Leipzig University, D-04107 Leipzig, Germany
| | - Peter F. Stadler
- Bioinformatics Group, Department of Computer Science, and Interdisciplinary Center for Bioinformatics, Leipzig University, D-04107 Leipzig, Germany
- Max Planck Institute for Mathematics in the Sciences, D-04103 Leipzig, Germany
- Institute for Theoretical Chemistry, University of Vienna, A-1090 Wien, Austria
- Facultad de Ciencias, Universidad National de Colombia, CO-111321 Bogotá, Colombia
- Santa Fe Insitute, Santa Fe, NM 87501, USA
| |
Collapse
|
4
|
An Approach to Identify Individual Functional Single Nucleotide Polymorphisms and Isoform MicroRNAs. BIOMED RESEARCH INTERNATIONAL 2019; 2019:6193673. [PMID: 31467902 PMCID: PMC6699389 DOI: 10.1155/2019/6193673] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 03/07/2019] [Revised: 05/28/2019] [Accepted: 06/24/2019] [Indexed: 01/08/2023]
Abstract
MicroRNAs (miRNAs) and single nucleotide polymorphisms (SNPs) play important roles in disease risk and development, especially cancer. Importantly, when SNPs are located in pre-miRNAs, they affect their splicing mechanism and change the function of miRNAs. To improve disease risk assessment, we propose an approach and developed a software tool, IsomiR_Find, to identify disease/phenotype-related SNPs and isomiRs in individuals. Our approach is based on the individual's samples, with SNP information extracted from the 1000 Genomes Project. SNPs were mapped to pre-miRNAs based on whole-genome coordinates and then SNP-pre-miRNA sequences were constructed. Moreover, we developed matpred2, a software tool to identify the four splicing sites of mature miRNAs. Using matpred2, we identified isomiRs and then verified them by searching within individual miRNA sequencing data. Our approach yielded biomarkers for biological experiments, mined functions of miRNAs and SNPs, improved disease risk assessment, and provided a way to achieve individualized precision medicine.
Collapse
|
5
|
Adaboost-SVM-based probability algorithm for the prediction of all mature miRNA sites based on structured-sequence features. Sci Rep 2019; 9:1521. [PMID: 30728425 PMCID: PMC6365589 DOI: 10.1038/s41598-018-38048-7] [Citation(s) in RCA: 6] [Impact Index Per Article: 1.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/26/2018] [Accepted: 12/18/2018] [Indexed: 02/07/2023] Open
Abstract
The significant role of microRNAs (miRNAs) in various biological processes and diseases has been widely studied and reported in recent years. Several computational methods associated with mature miRNA identification suffer various limitations involving canonical biological features extraction, class imbalance, and classifier performance. The proposed classifier, miRFinder, is an accurate alternative for the identification of mature miRNAs. The structured-sequence features were proposed to precisely extract miRNA biological features, and three algorithms were selected to obtain the canonical features based on the classifier performance. Moreover, the center of mass near distance training based on K-means was provided to improve the class imbalance problem. In particular, the AdaBoost-SVM algorithm was used to construct the classifier. The classifier training process focuses on incorrectly classified samples, and the integrated results use the common decision strategies of the weak classifier with different weights. In addition, the all mature miRNA sites were predicted by different classifiers based on the features of different sites. Compared with other methods, the performance of the classifiers has a high degree of efficacy for the identification of mature miRNAs. MiRFinder is freely available at https://github.com/wangying0128/miRFinder .
Collapse
|
6
|
Velandia-Huerto CA, Brown FD, Gittenberger A, Stadler PF, Bermúdez-Santana CI. Nonprotein-Coding RNAs as Regulators of Development in Tunicates. Results Probl Cell Differ 2018; 65:197-225. [PMID: 30083922 DOI: 10.1007/978-3-319-92486-1_11] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 06/08/2023]
Abstract
Tunicates, or urochordates, are a group of small marine organisms that are found widely throughout the seas of the world. As most plausible sister group of the vertebrates, they are of utmost importance for a comprehensive understanding of chordate evolution; hence, they have served as model organisms for many aspects of the developmental biology. Current genomic analysis of tunicates indicates that their genomes evolved with a fast rate not only at the level of nucleotide substitutions but also in terms of genomic organization. The latter involves genome reduction, rearrangements, as well as the loss of some important coding and noncoding RNA (ncRNAs) elements and even entire genomic regions that are otherwise well conserved. These observations are largely based on evidence from comparative genomics resulting from the analysis of well-studied gene families such as the Hox genes and their noncoding elements. In this chapter, the focus lies on the ncRNA complement of tunicates, with a particular emphasis on microRNAs, which have already been studied extensively for other animal clades. MicroRNAs are known as important regulators of key genes in animal development, and they are intimately related to the increase morphological complexity in higher metazoans. Here we review the discovery, evolution, and genome organization of the miRNA repertoire, which has been drastically reduced and restructured in tunicates compared to the chordate ancestor. Known functions of microRNAs as regulators of development in tunicates are a central topic. For instance, we consider the role of miRNAs as regulators of the muscle development and their importance in the regulation of the differential expression during the oral siphon regeneration. Beyond microRNAs, we touch upon the functions of some other ncRNAs such as yellow crescent RNA, moRNAs, RMST lncRNAs, or spliced-leader (SL) RNAs, which have diverse functions associated with the embryonic development, neurogenesis, and mediation of mRNA stability in general.
Collapse
Affiliation(s)
- Cristian A Velandia-Huerto
- Bioinformatics Group, Department of Computer Science, and Interdisciplinary Center for Bioinformatics, Universität Leipzig, Leipzig, Germany.
- Biology Department, Universidad Nacional de Colombia, Bogotá, Colombia.
| | - Federico D Brown
- Departamento de Zoologia, Instituto Biociências, Universidade de São Paulo, São Paulo, SP, Brazil
- Laboratorio de Biología del Desarrollo Evolutiva, Departamento de Ciencias Biológicas, Universidad de los Andes, Bogotá, Colombia
| | - Adriaan Gittenberger
- Institute of Biology, Leiden University, Leiden, Netherlands
- GiMaRIS, BioScience Park Leiden, Leiden, Netherlands
- Naturalis Biodiversity Center, Leiden, Netherlands
| | - Peter F Stadler
- Bioinformatics Group, Department of Computer Science, and Interdisciplinary Center for Bioinformatics, Universität Leipzig, Leipzig, Germany
| | | |
Collapse
|
7
|
Huang Y, Yang YB, Gao XC, Ren HT, Xiong JL, Sun XH. Genome-wide identification and characterization of microRNAs and target prediction by computational approaches in common carp. GENE REPORTS 2017. [DOI: 10.1016/j.genrep.2017.05.002] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/19/2022]
|
8
|
Spina EJ, Guzman E, Zhou H, Kosik KS, Smith WC. A microRNA-mRNA expression network during oral siphon regeneration in Ciona. Development 2017; 144:1787-1797. [PMID: 28432214 DOI: 10.1242/dev.144097] [Citation(s) in RCA: 15] [Impact Index Per Article: 2.1] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/25/2016] [Accepted: 04/10/2017] [Indexed: 12/14/2022]
Abstract
Here we present a parallel study of mRNA and microRNA expression during oral siphon (OS) regeneration in Ciona robusta, and the derived network of their interactions. In the process of identifying 248 mRNAs and 15 microRNAs as differentially expressed, we also identified 57 novel microRNAs, several of which are among the most highly differentially expressed. Analysis of functional categories identified enriched transcripts related to stress responses and apoptosis at the wound healing stage, signaling pathways including Wnt and TGFβ during early regrowth, and negative regulation of extracellular proteases in late stage regeneration. Consistent with the expression results, we found that inhibition of TGFβ signaling blocked OS regeneration. A correlation network was subsequently inferred for all predicted microRNA-mRNA target pairs expressed during regeneration. Network-based clustering associated transcripts into 22 non-overlapping groups, the functional analysis of which showed enrichment of stress response, signaling pathway and extracellular protease categories that could be related to specific microRNAs. Predicted targets of the miR-9 cluster suggest a role in regulating differentiation and the proliferative state of neural progenitors through regulation of the cytoskeleton and cell cycle.
Collapse
Affiliation(s)
- Elijah J Spina
- Department of Molecular, Cell and Developmental Biology, University of California, Santa Barbara, Santa Barbara, CA 93106, USA
| | - Elmer Guzman
- Department of Molecular, Cell and Developmental Biology, University of California, Santa Barbara, Santa Barbara, CA 93106, USA
| | - Hongjun Zhou
- Neuroscience Research Institute, University of California, Santa Barbara, Santa Barbara, CA 93106, USA
| | - Kenneth S Kosik
- Department of Molecular, Cell and Developmental Biology, University of California, Santa Barbara, Santa Barbara, CA 93106, USA.,Neuroscience Research Institute, University of California, Santa Barbara, Santa Barbara, CA 93106, USA
| | - William C Smith
- Department of Molecular, Cell and Developmental Biology, University of California, Santa Barbara, Santa Barbara, CA 93106, USA .,Neuroscience Research Institute, University of California, Santa Barbara, Santa Barbara, CA 93106, USA
| |
Collapse
|
9
|
Marques YB, de Paiva Oliveira A, Ribeiro Vasconcelos AT, Cerqueira FR. Mirnacle: machine learning with SMOTE and random forest for improving selectivity in pre-miRNA ab initio prediction. BMC Bioinformatics 2016; 17:474. [PMID: 28105918 PMCID: PMC5249014 DOI: 10.1186/s12859-016-1343-8] [Citation(s) in RCA: 14] [Impact Index Per Article: 1.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/31/2022] Open
Abstract
Background MicroRNAs (miRNAs) are key gene expression regulators in plants and animals. Therefore, miRNAs are involved in several biological processes, making the study of these molecules one of the most relevant topics of molecular biology nowadays. However, characterizing miRNAs in vivo is still a complex task. As a consequence, in silico methods have been developed to predict miRNA loci. A common ab initio strategy to find miRNAs in genomic data is to search for sequences that can fold into the typical hairpin structure of miRNA precursors (pre-miRNAs). The current ab initio approaches, however, have selectivity issues, i.e., a high number of false positives is reported, which can lead to laborious and costly attempts to provide biological validation. This study presents an extension of the ab initio method miRNAFold, with the aim of improving selectivity through machine learning techniques, namely, random forest combined with the SMOTE procedure that copes with imbalance datasets. Results By comparing our method, termed Mirnacle, with other important approaches in the literature, we demonstrate that Mirnacle substantially improves selectivity without compromising sensitivity. For the three datasets used in our experiments, our method achieved at least 97% of sensitivity and could deliver a two-fold, 20-fold, and 6-fold increase in selectivity, respectively, compared with the best results of current computational tools. Conclusions The extension of miRNAFold by the introduction of machine learning techniques, significantly increases selectivity in pre-miRNA ab initio prediction, which optimally contributes to advanced studies on miRNAs, as the need of biological validations is diminished. Hopefully, new research, such as studies of severe diseases caused by miRNA malfunction, will benefit from the proposed computational tool.
Collapse
Affiliation(s)
- Yuri Bento Marques
- Department of Informatics, Universidade Federal de Viçosa, Viçosa, 36570-900, Brazil.,Instituto Federal do Norte de Minas, Rua Mocambi, Teófilo Otoni, 39800-430, Brazil
| | - Alcione de Paiva Oliveira
- Department of Informatics, Universidade Federal de Viçosa, Viçosa, 36570-900, Brazil.,Department of Computer Science, University of Sheffield, Western Bank S10 2TNSheffield, UK
| | | | | |
Collapse
|
10
|
Improving classification of mature microRNA by solving class imbalance problem. Sci Rep 2016; 6:25941. [PMID: 27181057 PMCID: PMC4867574 DOI: 10.1038/srep25941] [Citation(s) in RCA: 5] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/18/2016] [Accepted: 04/22/2016] [Indexed: 11/29/2022] Open
Abstract
MicroRNAs (miRNAs) are ~20–25 nucleotides non-coding RNAs, which regulated gene expression in the post-transcriptional level. The accurate rate of identifying the start sit of mature miRNA from a given pre-miRNA remains lower. It is noting that the mature miRNA prediction is a class-imbalanced problem which also leads to the unsatisfactory performance of these methods. We improved the prediction accuracy of classifier using balanced datasets and presented MatFind which is used for identifying 5′ mature miRNAs candidates from their pre-miRNA based on ensemble SVM classifiers with idea of adaboost. Firstly, the balanced-dataset was extract based on K-nearest neighbor algorithm. Secondly, the multiple SVM classifiers were trained in orderly using the balance datasets base on represented features. At last, all SVM classifiers were combined together to form the ensemble classifier. Our results on independent testing dataset show that the proposed method is more efficient than one without treating class imbalance problem. Moreover, MatFind achieves much higher classification accuracy than other three approaches. The ensemble SVM classifiers and balanced-datasets can solve the class-imbalanced problem, as well as improve performance of classifier for mature miRNA identification. MatFind is an accurate and fast method for 5′ mature miRNA identification.
Collapse
|
11
|
MatPred: Computational Identification of Mature MicroRNAs within Novel Pre-MicroRNAs. BIOMED RESEARCH INTERNATIONAL 2015; 2015:546763. [PMID: 26682221 PMCID: PMC4670854 DOI: 10.1155/2015/546763] [Citation(s) in RCA: 5] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 07/26/2015] [Revised: 09/18/2015] [Accepted: 09/28/2015] [Indexed: 12/22/2022]
Abstract
Background. MicroRNAs (miRNAs) are short noncoding RNAs integral for regulating gene expression at the posttranscriptional level. However, experimental methods often fall short in finding miRNAs expressed at low levels or in specific tissues. While several computational methods have been developed for predicting the localization of mature miRNAs within the precursor transcript, the prediction accuracy requires significant improvement. Methodology/Principal Findings. Here, we present MatPred, which predicts mature miRNA candidates within novel pre-miRNA transcripts. In addition to the relative locus of the mature miRNA within the pre-miRNA hairpin loop and minimum free energy, we innovatively integrated features that describe the nucleotide-specific RNA secondary structure characteristics. In total, 94 features were extracted from the mature miRNA loci and flanking regions. The model was trained based on a radial basis function kernel/support vector machine (RBF/SVM). Our method can predict precise locations of mature miRNAs, as affirmed by experimentally verified human pre-miRNAs or pre-miRNAs candidates, thus achieving a significant advantage over existing methods. Conclusions. MatPred is a highly effective method for identifying mature miRNAs within novel pre-miRNA transcripts. Our model significantly outperformed three other widely used existing methods. Such processing prediction methods may provide important insight into miRNA biogenesis.
Collapse
|
12
|
miReader: Discovering Novel miRNAs in Species without Sequenced Genome. PLoS One 2013; 8:e66857. [PMID: 23805282 PMCID: PMC3689854 DOI: 10.1371/journal.pone.0066857] [Citation(s) in RCA: 22] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/14/2013] [Accepted: 05/11/2013] [Indexed: 12/26/2022] Open
Abstract
Along with computational approaches, NGS led technologies have caused a major impact upon the discoveries made in the area of miRNA biology, including novel miRNAs identification. However, to this date all microRNA discovery tools compulsorily depend upon the availability of reference or genomic sequences. Here, for the first time a novel approach, miReader, has been introduced which could discover novel miRNAs without any dependence upon genomic/reference sequences. The approach used NGS read data to build highly accurate miRNA models, molded through a Multi-boosting algorithm with Best-First Tree as its base classifier. It was comprehensively tested over large amount of experimental data from wide range of species including human, plants, nematode, zebrafish and fruit fly, performing consistently with >90% accuracy. Using the same tool over Illumina read data for Miscanthus, a plant whose genome is not sequenced; the study reported 21 novel mature miRNA duplex candidates. Considering the fact that miRNA discovery requires handling of high throughput data, the entire approach has been implemented in a standalone parallel architecture. This work is expected to cause a positive impact over the area of miRNA discovery in majority of species, where genomic sequence availability would not be a compulsion any more.
Collapse
|