DOCSLIB.ORG

Computational Methods for Predicting Protein-Protein Interactions and Binding Sites

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Computational Methods for Predicting Protein-Protein Interactions and Binding Sites Western University Scholarship@Western Electronic Thesis and Dissertation Repository 8-24-2020 11:30 AM Computational Methods for Predicting Protein-protein Interactions and Binding Sites Yiwei Li, The University of Western Ontario Supervisor: Ilie, Lucian, The University of Western Ontario A thesis submitted in partial fulfillment of the equirr ements for the Doctor of Philosophy degree in Computer Science © Yiwei Li 2020 Follow this and additional works at: https://ir.lib.uwo.ca/etd Part of the Bioinformatics Commons, and the Other Computer Sciences Commons Recommended Citation Li, Yiwei, "Computational Methods for Predicting Protein-protein Interactions and Binding Sites" (2020). Electronic Thesis and Dissertation Repository. 7182. https://ir.lib.uwo.ca/etd/7182 This Dissertation/Thesis is brought to you for free and open access by Scholarship@Western. It has been accepted for inclusion in Electronic Thesis and Dissertation Repository by an authorized administrator of Scholarship@Western. For more information, please contact [email protected]. Abstract Proteins are essential to organisms and participate in virtually every process within cells. Quite often, they keep the cells functioning by interacting with other proteins. This process is called protein-protein interaction (PPI). The bonding amino acid residues during the process of protein-protein interactions are called PPI binding sites. Identifying PPIs and PPI binding sites are fundamental problems in system biology. Experimental methods for solving these two problems are slow and expensive. There- fore, great efforts are being made towards increasing the performance of computational methods. We present DELPHI, a deep learning based program for PPI site prediction and SPRINT, an algorithmic based program for PPI prediction. Both programs have been compared to the state-of-the-art programs on several datasets. Both DELPHI and SPRINT are more accurate than the competing method. SPRINT is also orders of magnitudes faster while using very little memory. The dataset and source code for both DELPHI and SPRINT are publicly available at: github.com/lucian-ilie and and www.csd.uwo.ca/~ilie/software.html Keywords: Bioinformatics, SPRINT, DELPHI, Protein-protein interaction, deep learning, Protein-protein interaction prediction, Protein-protein interaction binding sites prediction i Lay Summary Proteins are essential to organisms and participate in virtually every process within cells. Quite often, they keep the cells functioning by interacting with other proteins. This process is called protein-protein interaction (PPI). The bonding amino acid residues during the process of protein-protein interactions are called PPI binding sites. Identifying PPIs and PPI binding sites are fundamental problems in system biology. Experimental methods for solving these two problems are slow and expensive. There- fore, great efforts are being made towards increasing the performance of computational methods. We present two computational methods: DELPHI, for PPI site prediction and SPRINT, for PPI prediction. Both programs surpass the state-of-the-art programs and they are freely available at www.csd.uwo.ca/~ilie/software.html ii Acknowledgements First and foremost, my sincere gratitude goes to my supervisor, Dr. Lucian Ilie. He has not only taught me knowledge in computer science, more importantly, the right way of doing things in general. My growth throughout my master and PhD period would not be possible without his inspiring lectures, hands-on instructions and manuscript editing, countless video discussions, and mental supports. I would also like to express my heartfelt appreciation to my parents. To me, leaving home and pursing new life is exciting, adventures, and rewarding, but to them, in addi- tion, is not being able to spend holidays with their only kid. I gradually understand their feelings these years, and I would like to say: thank you and I love you Dad and Mom. I would also like to thank my fiancee Karen Qi and her parents. Studying and working at the same time is way harder than I initially though, but you made it easier. Toronto feels like home because of you. I thank my lab mates, Shaofeng Jiang, Fang Han, Qin Dong, Zhewei Liang, Nilesh Khiste, Yi Liu, Weiping Sun, Ehsan Haghshenas, Mike Molnar in Middlesex College 222, Computer Science Department, Western University. I will certainly remember the great hikes, BBQs, trips, gatherings we had together as colleagues and friends. My appreciation also goes to Huawei Toronto Research Center. I am horned to be in such a great team with great talents. The engineering skills I learned at Huawei benefit me a lot in my research projects. To all my teammates on the Table Tennis Varsity Team, Western University: table tennis is an important part of me, and one of the reasons is the amazing people I met, like you. iii Contents Abstract i Lay Summary ii Acknowlegements iii List of Figures viii List of Tables xi 1 Introduction 1 1.1 DNA . 1 1.2 Protein . 1 1.3 Protein-protein Interaction Prediction . 4 1.4 Protein-protein Interaction Binding Sites Prediction . 6 1.5 Thesis Overview . 9 2 DELPHI 10 2.1 Background . 10 2.1.1 Deep Learning in Bioinformatics . 10 2.1.2 Basic Notions and Definitions . 14 Deep Neural Networks . 14 Training . 16 Inference . 18 Convolutional Neural Networks . 18 Recurrent Neural Networks . 22 iv Ensemble Networks . 24 Dropout Layers . 26 Training, Validation, and Testing Dataset . 27 Data Augmentations and Sampling . 27 2.1.3 Previous Methods . 28 PIPE-sites . 28 DLPred . 29 DeepPPISP . 30 SCRIBER . 31 2.2 Methods . 32 2.2.1 Training Data Preparation . 33 Raw Training Data . 33 Similarities Eliminations . 33 Data Split . 34 2.2.2 Features . 34 2.2.3 Model Architecture . 39 Architecture Overview . 39 Many-to-one Structure . 39 Architecture of the CNN Network . 41 Architecture of the RNN Network . 42 Architecture of the Ensemble Network . 42 2.2.4 Parameter/Hyper-parameter Tuning . 43 2.2.5 Implementation . 44 Environment Configuration . 45 Class Weights . 46 Data Shuffling . 46 2.2.6 The DELPHI Web Server . 47 The Architecture of the Web Server . 47 Front End Server Configuration . 48 Back End Server Configuration . 49 v Communications between the Front and Back End Servers . 49 Pre-computing PSSMs . 50 Job Scheduling . 50 2.3 Results . 50 2.3.1 Testing Datasets . 51 2.3.2 Evaluation Scheme . 53 2.3.3 Performance Comparison on Dset 448 and Dset 355 . 54 2.3.4 Performance Comparison on Dset 186, Dset 164, and Dset 72 . 54 2.3.5 Ablation Study . 56 Feature Evaluation . 56 The Evaluation of the Model Architecture and the Novel Features 57 2.3.6 Evolutionary Conservation...................... 58 2.3.7 Accuracy of PBR Prediction..................... 59 2.3.8 Human Proteome Prediction..................... 62 2.3.9 Availability . 62 2.4 Conclusion . 62 3 SPRINT 64 3.1 Background . 64 3.1.1 Similarity Search . 64 BLAST Seeds . 65 Spaced Seeds . 67 Multiple Spaced Seeds . 68 Substitution Matrix . 69 Interactome Prediction . 70 3.1.2 Previous Methods . 70 PIPE . 73 Martin's Program . 76 Shen's Program . 77 Guo's Program . 78 vi Ding's Program . 80 3.2 Methods . 80 3.2.1 Basic Idea . 81 3.2.2 Detecting Similarities . 83 3.2.3 Predicting Interactions . 88 Post-processing Similarities . 88 Scoring Function . 88 3.2.4 Implementations . 90 Tuneable Parameters . 90 Pseudocode . 90 System Configuration . 92 3.3 Results . 92 3.3.1 Datasets Classification . 92 3.3.2 Datasets . 92 3.3.3 Competing Methods . 94 3.3.4 Comparative Analysis on Park & Marcotte's Datasets . 94 3.3.5 Comparative Analysis on Seven Human Datasets . 95 3.3.6 Comparative Analysis on Human Interactome Prediction . 97 3.3.7 Availability . 100 3.4 Conclusion . 100 4 Conclusion and Future Research 102 4.1 Common Deep Learning Practises in Bioinformatics . 102 4.1.1 Data Preparation . 103 Comparative Analysis . 104 Improving Results . 104 4.2 Future Research . 105 Bibliography 106 Curriculum Vitae 124 vii List of Figures 1.1 Illustration of transcription and Translation . 2 1.2 The primary structure of a protein . 3 1.3 The primary, secondary, tertiary, and quaternary structure of proteins . 3 1.4 A FASTA file example . 4 1.5 Protein complex . 5 1.6 A human protein-protein interaction network . ..
Load more

Recommended publications
  • ISMB 2008 Toronto
    ISMB 2008 Toronto The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters Citation Linial, Michal, Jill P. Mesirov, B. J. Morrison McKay, and Burkhard Rost. 2008. ISMB 2008 Toronto. PLoS Computational Biology 4(6): e1000094. Published Version doi:10.1371/journal.pcbi.1000094 Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:11213310 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA Message from ISCB ISMB 2008 Toronto Michal Linial1,2, Jill P. Mesirov1,3, BJ Morrison McKay1*, Burkhard Rost1,4 1 International Society for Computational Biology (ISCB), University of California San Diego, La Jolla, California, United States of America, 2 Sudarsky Center, The Hebrew University of Jerusalem, Jerusalem, Israel, 3 Broad Institute of MIT and Harvard, Cambridge, Massachusetts, United States of America, 4 Department of Biochemistry and Molecular Biophysics, Columbia University, New York, New York, United States of America the integration of students, and for the of ISMB. One meeting in South Asia support of young leaders in the field. (InCoB; http://incob.binfo.org.tw/) has ISMB has also become a forum for already been sponsored by ISCB, and reviewing the state of the art in the many another one in North Asia is going to fields of this growing discipline, for follow. ISMB itself has also been held in introducing new directions, and for an- Australia (2003) and Brazil (2006).
    [Show full text]
  • From DNA Sequence to Chromatin Dynamics: Computational Analysis of Transcriptional Regulation
    From DNA Sequence to Chromatin Dynamics: Computational Analysis of Transcriptional Regulation Thesis submitted for the degree of “Doctor of Philosophy” by Tommy Kaplan Submitted to the Senate of the Hebrew University May 2008 This work was carried out under the supervision of Prof. Nir Friedman and Prof. Hanah Margalit Abstract All cells of a living organism share the same DNA. Yet, they differ in structure, activities and interactions. These differences arise through a tight regulatory system which activates different genes and pathways to fit the cell’s specialization, condition, and requirements. Deciphering the regulatory mechanisms underlying a living cell is one of the fundamental challenges in biology. Such knowledge will allow us to better understand how cells work, how they respond to external stimuli, what goes wrong in diseases like cancer (which often involves disruption of gene regulation), and how it can be fought. In my PhD, I focus on regulation of gene expression from three perspectives. First, I present an innovative algorithm for identifying the target genes of novel transcription factors, based on their protein sequence (Chapter 1). Second, I consider how several transcription factors cooperate to process external stimuli and alter the behavior of the cell (Chapter 2). Finally, I study how the genomic position of nucleosomes and their covalent modifications modulate the accessibility of DNA to transcription factors, thus adding a fascinating dimension to transcriptional regulation (Chapters 3 and 4). To understand transcriptional regulation, one should first reconstruct the architecture of the cell’s regulatory map, thus identifying which genes are regulated by which transcription factors (TFs).
    [Show full text]
  • A Hidden Markov Model Framewrok for Studying Regulation from Chromatin Through RNA
    Submitted in part fulfilment of the requirements for the degree of Master of Science in Computer Science and Computational Biology A Hidden Markov Model Framewrok For Studying Regulation From Chromatin Through RNA Eran Rosenthal Supervisor: Dr. Tommy Kaplan October 2017 Hebrew University of Jerusalem Faculty of Science - Computer Science and Engineering Acknowledgments I would like to deeply thank Tommy, for being a great supervisor, giving very useful advices and scientific tutoring. Tommy has great ability to express complex concepts in simple, illustrative stories, and he introduced me to different kinds of experimental data and computational approaches. I would like also to thank for the members of the lab, for the interesting ideas and discussions. I would like to thank for all the people I had worked in collaboration in my master: • Moshe Oren and Gilad Fuchs for the providing me their experimental data of nascent RNA with RNF20 knockdown for studying the role of monoubiquity- lation of H2B. • Michael Berger and Yuval Malka for the exciting project of studying post- transcriptional 3’UTR cleavage of mRNA transcripts, and for Hanah Margalit and Avital Shimony who helped us in the analysis of miRNA regulation in this project. i Abstract Information flows from DNA to RNA, through transcription, followed by trans- lation of the RNA transcript to protein. A viable cell must have proper regulation on genes expression. There are varieties of regulation mechanisms along the way, which include regulation on the DNA and expression level of genes, post-transcriptional regulation, and post-translation regulation. Most of the cells in our body share the same DNA, but they have different func- tions.
    [Show full text]
  • EYAL AKIVA, Phd
    Eyal Akiva CV, Nov. 2017 EYAL AKIVA, PhD Department of Bioengineering and Therapeutic Sciences Phone +1-650-504-9008 University of California at San Francisco Email [email protected] 1700 4th street, San Francisco, Web www.babbittlab.ucsf.edu/eakiva CA, USA EDUCATION 2012-2017 Post-doctoral fellowship at UCSF, Dept. Of Bioengineering and Therapeutic Sciences. Host: Prof. Patricia Babbitt. 2010-2012 Post-doctoral fellowship at UCSF, Dept. Of Bioengineering and Therapeutic Sciences. Host: Prof. Tanja Kortemme. 2004-2010 PhD at The Hebrew University of Jerusalem (Israel), bioinformatics. Host: Prof. Hanah Margalit. “Various Aspects of Modularity in Protein-Protein Interaction". 2001-2004 MSc at The Hebrew University of Jerusalem (Israel), bioinformatics and human genetics. Host: Prof. Muli Ben-Sasson. “Exploiting the Exploiters: Identification of Virus-Host Pep- tide Mimicry as a Source for Modules of Functional Significance”. MAGNA CUM LAUDE. 1997-2000 BSc at Bar-Ilan University (Israel), biology (major) and computer science (minor). Final project advisor: Prof. Ramit Mehr “Modeling the Evolution of the Immune System: a Sim- ulation of the Evolution of Genes that Encode the Variable Regions of Immunoglobulins”. MAGNA CUM LAUDE. 1996-1997 First year of "Industrial Engineering and Management" studies, Tel-Aviv University, Israel. OTHER WORK EXPERIENCE 2000-01 ‘Do-coop technologies’: Team leader and chemistry/microbiology researcher; development of biological applications and manufacture of proprietary nanoparticles (Or Yehuda, Israel and Tel-Aviv University (Prof. Eshel Ben-Jacob’s lab at the school of physics)). FUNDING, HONORS AND AWARDS 2017 Grant: Co-PI, “Utilizing metagenomic sequences for enzyme function prediction”, Joint Genome Institute (US Department of Energy) (http://jgi.doe.gov/doe-user-facilities-ficus- join-forces-to-tackle-biology-big-data/).
    [Show full text]
  • Speakers Info
    CAJAL Online Lecture Series Single Cell Transcriptomics November 2nd-6th, 2020 Keynote speakers Naomi HABIB, PhD | Edmond & Lily Safra Center for Brain Sciences (ELSC), Israel Naomi Habib is an assistant professor at the ELSC Brain Center at the Hebrew University of Jerusalem since July 2018. Habib's research focuses on understanding how complex interactions between diverse cell types in the brain and between the brain and other systems in the body, are mediating neurodegenerative diseases and other aging-related pathologies. Naomi combines in her work computational biology, genomics and genome-engineering, and is a pioneer in single nucleus RNA-sequencing technologies and their applications to study cellular diversity and molecular processes in the brain. Naomi did her postdoctoral at the Broad Institute of MIT/Harvard working with Dr. Feng Zhang and Dr. Aviv Regev, and earned her PhD in computational biology from the Hebrew University of Jerusalem in Israel, working with Prof. Nir Friedman and Prof. Hanah Margalit. Selected publications: - Habib N, McCabe C*, Medina S*, Varshavsky M*, Kitsberg D, Dvir R, Green G, Dionne D, Nguyen L, Marshall J.L, Chen F, Zhang F, Kaplan T, Regev A, Schwartz M. (2019) Unique disease-associated astrocytes in Alzheimer’s disease. Nature Neuroscience. In Press. - Habib N*, Basu A*, Avraham-Davidi I*, Burks T, Choudhury SR, Aguet F, Gelfand E, Ardlie K, Weitz DA, Rozenblatt-Rosen O, Zhang F, and Regev A. (2017). Deciphering cell types in human archived brain tissues by massively-parallel single nucleus RNA-seq. Nature Methods. Oct;14(10):955-958. - Habib N*, Li Y*, Heidenreich M, Sweich L, Avraham-Davidi I, Trombetta J, Hession C, Zhang F, Regev A.
    [Show full text]
  • I S C B N E W S L E T T
    ISCB NEWSLETTER FOCUS ISSUE {contents} President’s Letter 2 Member Involvement Encouraged Register for ISMB 2002 3 Registration and Tutorial Update Host ISMB 2004 or 2005 3 David Baker 4 2002 Overton Prize Recipient Overton Endowment 4 ISMB 2002 Committees 4 ISMB 2002 Opportunities 5 Sponsor and Exhibitor Benefits Best Paper Award by SGI 5 ISMB 2002 SIGs 6 New Program for 2002 ISMB Goes Down Under 7 Planning Underway for 2003 Hot Jobs! Top Companies! 8 ISMB 2002 Job Fair ISCB Board Nominations 8 Bioinformatics Pioneers 9 ISMB 2002 Keynote Speakers Invited Editorial 10 Anna Tramontano: Bioinformatics in Europe Software Recommendations11 ISCB Software Statement volume 5. issue 2. summer 2002 Community Development 12 ISCB’s Regional Affiliates Program ISCB Staff Introduction 12 Fellowship Recipients 13 Awardees at RECOMB 2002 Events and Opportunities 14 Bioinformatics events world wide INTERNATIONAL SOCIETY FOR COMPUTATIONAL BIOLOGY A NOTE FROM ISCB PRESIDENT This newsletter is packed with information on development and dissemination of bioinfor- the ISMB2002 conference. With over 200 matics. Issues arise from recommendations paper submissions and over 500 poster submis- made by the Society’s committees, Board of sions, the conference promises to be a scientific Directors, and membership at large. Important feast. On behalf of the ISCB’s Directors, staff, issues are defined as motions and are discussed EXECUTIVE COMMITTEE and membership, I would like to thank the by the Board of Directors on a bi-monthly Philip E. Bourne, Ph.D., President organizing committee, local organizing com- teleconference. Motions that pass are enacted Michael Gribskov, Ph.D., mittee, and program committee for their hard by the Executive Committee which also serves Vice President work preparing for the conference.
    [Show full text]
  • Study of Seed-Based Microrna Targeting in Mammals
    TECHNISCHE UNIVERSITAT¨ MUNCHEN¨ Lehrstuhl f¨urGenomorientierte Bioinformatik Study of seed-based microRNA targeting in mammals Florian A. B¨uttner Vollst¨andigerAbdruck der von der Fakult¨atWissenschaftszentrum Weihenstephan f¨urErn¨ahrung,Landnutzung und Umwelt der Technischen Universit¨atM¨unchen zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigten Dissertation. Vorsitzender: Univ.-Prof. Dr. B. K¨uster Pr¨uferder Dissertation: 1. Univ.-Prof. Dr. H.-W. Mewes 2. Univ.-Prof. Dr. D. Frischmann Die Dissertation wurde am 07.08.2013 bei der Technischen Universit¨atM¨unchen eingereicht und durch die Fakult¨atWissenschaftszentrum Weihenstephan f¨urEr- n¨ahrung,Landnutzung und Umwelt am 22.11.2013 angenommen. To Katrin and Paulo iii Danksagung Zuerst m¨ochte ich mich bei meinem Doktorvater Prof. Dr. Hans-Werner Mewes bedanken, mir diese Doktorarbeit erm¨oglicht zu haben, sowie f¨urdie Betreuung und wohlwollende Unterst¨utzungin den letzten Jahren. F¨urstimulierende und spannende Zusammenarbeit und Diskussionen m¨ochte ich mich vor allem bei Daniel Ellwanger und Florian Giesert, sowie bei Sebastian Toepel, Matthias Arnold, Dominik Lutter und J¨ornLeonhardt bedanken. Weiter geht mein Dank an alle Kollegen vom Institut f¨urBioinformatik und Systembiologie und vom Lehrstuhl f¨urGenomorientierte Bioinformatik in Wei- henstephan f¨urdas freundliche, kollegiale und unterhaltsame Miteinander. Erw¨ahnung finden m¨ussenan dieser Stelle auch meine Freunde und Bekannten außerhalb von IBIS und Universit¨at.Die w¨ochentlichen
    [Show full text]
  • Interact: Interaction Network Inference from Vector
    INtERAcT: Interaction Network Inference from Vector Representations of Words Matteo Manica1,2,*, Roland Mathis1,*, Mar´ıaRodr´ıguezMart´ınez1, y ftte,lth,[email protected] 1 IBM Research Z¨urich 2 ETH - Z¨urich * Shared first authorship y Corresponding author Abstract Motivation In recent years, the number of biomedical publications has steadfastly grown, resulting in a rich source of untapped new knowledge. Most biomedical facts are however not readily available, but buried in the form of unstructured text, and hence their exploitation requires the time-consuming manual curation of published articles. Here we present INtERAcT, a novel approach to extract protein-protein interactions from a corpus of biomedical articles related to a broad range of scientific domains in a completely unsupervised way. INtERAcT exploits vector representation of words, computed on a corpus of domain specific knowledge, and implements a new metric that estimates an interaction score between two molecules in the space where the corresponding words are embedded. Results We demonstrate the power of INtERAcT by reconstructing the molecular pathways asso- ciated to 10 different cancer types using a corpus of disease-specific articles for each cancer type. We evaluate INtERAcT using STRING database as a benchmark, and show that our metric outperforms currently adopted approaches for similarity computation at the task of identifying known molecular interactions in all studied cancer types. Furthermore, our approach does not require text annotation, manual curation or the definition of semantic rules based on expert knowledge, and hence it can be easily and efficiently applied to dif- ferent scientific domains. Our findings suggest that INtERAcT may increase our capability to summarize the understanding of a specific disease using the published literature in an automated and completely unsupervised fashion.
    [Show full text]
  • Download Flyer
    COLD SPRING HARBOR ASIA ORGANIZERS(Speaker, Affiliation, Country/Region） Steven E. Brenner Frontiers in University of California, Berkeley, USA A Keith Dunker Indiana University School of Medicine, USA Computational Biology Julian Gough MRC Laboratory of Molecular Biology, UK Suzhou, China September 3-7, 2018 Luhua Lai & Bioinformatics Peking University, China Yunlong Liu Abstract deadline: July 13, 2018 Indiana University School of Medicine, USA MAJOR TOPICS Precision medicine, human genome variation, disease & diagnosis Molecular evolution Pathways, networks & developmental biology Molecular structure, with pioneering techniques Molecular machines, their functions & dynamics Intrinsically disordered proteins & their functions RNA function, regulation & splicing 3D genomics & regulatory inferences Single cell analysis KEYNOTE SPEAKERS (Speaker, Affiliation, Country/Region） Nancy Cox, Vanderbilt University, USA Yoshihide Hayashizaki, RIKEN Research Cluster for Innovation, JAPAN INVITED SPEAKERS (Speaker, Affiliation, Country/Region） Russ Altman, Stanford University, USA Manolis Kellis, MIT Computer Science and Broad Institute, USA Lukasz Kurgan, Virginia Commonwealth University, USA Peer Bork, European Molecular Biology Laboratory, GERMANY Luhua Lai, Peking University, CHINA Steven Brenner, University of California, Berkeley, USA Michal Linial, The Hebrew University of Jerusalem, ISRAEL Angela Brooks, University of California, Santa Cruz, USA Yunlong Liu, Indiana University School of Medicine, USA Luonan Chen, Shanghai Institutes for
    [Show full text]
  • Transformer Neural Networks for Protein Prediction Tasks
    bioRxiv preprint doi: https://doi.org/10.1101/2020.06.15.153643; this version posted June 16, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. TRANSFORMING THE LANGUAGE OF LIFE:TRANSFORMER NEURAL NETWORKS FOR PROTEIN PREDICTION TASKS Ananthan Nambiar ∗ Maeve Heflin∗ Department of Bioengineering Department of Computer Science Carl R. Woese Inst. for Genomic Biol. Carl R. Woese Inst. for Genomic Biol. University of Illinois at Urbana-Champaign University of Illinois at Urbana-Champaign Urbana, IL 61801 Urbana, IL 61801 [email protected] Simon Liu∗ Sergei Maslov Department of Computer Science Department Bioengineering Carl R. Woese Inst. for Genomic Biol. Department of Physics University of Illinois at Urbana-Champaign Carl R. Woese Inst. for Genomic Biol. Urbana, IL 61801 University of Illinois at Urbana-Champaign Urbana, IL 61801 Mark Hopkinsy Anna Ritzy Department of Computer Science Department of Biology Reed College Reed College Portland, OR 97202 Portland, OR 97202 June 16, 2020 ABSTRACT The scientific community is rapidly generating protein sequence information, but only a fraction of these proteins can be experimentally characterized. While promising deep learning approaches for protein prediction tasks have emerged, they have computational limitations or are designed to solve a specific task. We present a Transformer neural network that pre-trains task-agnostic sequence representations. This model is fine-tuned to solve two different protein prediction tasks: protein family classification and protein interaction prediction.
    [Show full text]
  • Thesis (6.404Mb)
    From DEPARTMENT OF MEDICINE, SOLNA Karolinska Institutet, Stockholm, Sweden UNDERSTANDING THE ROLE OF NON-CODING RNAs IN SKIN HOMEOSTASIS AND CANCER KUNAL DAS MAHAPATRA Stockholm, 2020 All previously published papers were reproduced with permission from the publisher. Published by Karolinska Institutet. Printed by Eprint AB, 2020 © Kunal Das Mahapatra, 2020 ISBN 978-91-7831-900-8 Understanding the role of non-coding RNAs in skin homeostasis and cancer THESIS FOR DOCTORAL DEGREE (Ph.D.) The thesis will be defended on August 28, 2020 Location: Visiongatan 18, CMM Lecture Hall, Ground Floor, L8:00 Center for Molecular Medicine, Solna, Stockholm By KUNAL DAS MAHAPATRA Principal Supervisor: Opponent: Andor Pivarcsi, Associate Professor Wim Declercq, Professor Karolinska Institutet Ghent University Department of Medicine, Solna Department of Biomedical Molecular Biology Division of Dermatology and Venereology Examination Board: Co-supervisor(s): Gonçalo Castelo-Branco, Associate Professor Mona Ståhle, Professor Karolinska Institutet Karolinska Institutet Department of Medical Biochemistry and Biophysics Department of Medicine, Solna Division of Dermatology and Venereology Hans Törmä, Professor Uppsala University Leonard Girnita, Associate Professor Department of Dermatology and Venereology Karolinska Institutet Department of Oncology-Pathology Elisabet Nylander, Professor Umeå University Johan Hansson, Professor Department of Public Health and Clinical Medicine Karolinska Institutet Department of Oncology-Pathology ABSTRACT The epidermis is a stratified epithelium with continuous self-renewing capacity. As the outermost layer of our body, it provides a protective barrier against external trauma, produces pigmentation, and keeps the skin hydrated. Keratinocytes are the primary constituent cells type within the epidermis. A fine balance is maintained between keratinocyte cell proliferation and differentiation to sustain a functional epidermis.
    [Show full text]
  • Network-Level Study of Protein-Protein Interactions
    NETWORK-LEVEL STUDY OF PROTEIN-PROTEIN INTERACTIONS: ANALYSIS AND PREDICTION by Lei Huang A dissertation submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Computer Science Spring 2017 c 2017 Lei Huang All Rights Reserved NETWORK-LEVEL STUDY OF PROTEIN-PROTEIN INTERACTIONS: ANALYSIS AND PREDICTION by Lei Huang Approved: Kathleen F. McCoy, Ph.D. Chair of the Department of Computer and Information Sciences Approved: Babatunde Ogunnaike, Ph.D. Dean of the College of Engineering Approved: Ann L. Ardis, Ph.D. Senior Vice Provost for Graduate and Professional Education I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy. Signed: Li Liao, Ph.D. Professor in charge of dissertation I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy. Signed: Cathy H. Wu, Ph.D. Professor in charge of dissertation I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy. Signed: Vijay K. Shanker, Ph.D. Member of dissertation committee I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.
    [Show full text]