Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Interdisciplinary research collaboration to unravel new insight on hematological malignancies Merja Heinäniemi, PhD Academy Research Fellow, Adj. Prof Institute of Biomedicine, School of Medicine The problem t(15;17) translocation in AML Zhu et al. Blood 2001 Systems genomics research group • We approach problems from a systems view point • With an interdisciplinary skill set Biochemistry Genomics Biomedicine Computer science Systems biology So what goes wrong in AML? A complex dynamic system in its simplest form Molecular interaction networks are central for cell function Decisions & memory Principles from theoretical physics A cell with two regulatory molecules Molecular interaction networks are central for cell function Decisions & memory Principles from theoretical physics Regulatory molecules control the system state Models of cell type dynamics were introduced by Prof. Stuart Kauffman and a lot of experimental validation came from Prof. Sui Huang – two brilliant scientists holding an MD degree who started applying theoretical physics in their work Decisions & memory Principles from theoretical physics Prof Ilya Shmulevich (left, Institute for Systems Biology) and Prof Matti Nykter (below, University of Tampere) introduced me to computational biology Heinäniemi et al Nature Methods 2013 Together we discovered key regulators of normal cell differentiation from data Heinäniemi et al Nature Methods 2013 Coming back to our problem: Regulatory molecules dysfunction in cancer Transitions between cellular states depend on key regulatory proteins Zhu et al. Blood 2001 t(15;17) translocation in AML Can we observe and quantify from data what happens in hematological malignancies? - Regulators of normal cell differentiation lost Differentiation blocked manuscript in preparation Can we observe and quantify from data what happens in hematological malignancies? Data-driven visualization and clustering of patient data - Identification of disease-specific regulators Summary 1 • A theoretical framework exists for studying complex dynamical systems which can be applied to biomedical questions • Biological systems are dynamic and complex: diseases can be understood as network perturbations and state transitions The problem – Biology – Theoretical framework The problem – Biology – Theoretical framework - Data How do we study this ON A PRACTICAL LEVEL A lot of our daily work deals with large datasets and therefore requires bioinformatics gene expression level Traditionally research has focused on measurements quantifying specific genes and proteins from samples Omics technologies and data availability have changed biomedicine Input from computational fields in data analysis Interdisciplinarity Time in wet-lab is only small part of research projects What has been measured • Availability and ease of use of the technologies matters: many DNA ja RNA level measurements (gene expresssion levels, genetic variants) • Less data on proteins and metabolites (so far) Relevant information is obtained from various measurement types – many large data tables! ANALYSIS Data integration: current challenge http://cancergenome.nih.gov/ Summary 2 • Omics technologies are available and generate large volumes of data • This has changed the way biomedical research is conducted: we need new skills! • Data matrices contain 104-105 measurement values from each sample and many such results need to be integrated Data on hematological malignancies 10K transcriptomes were collected and curated Can we diagnose patients based on molecular profiles? What can we learn about these cancers? Challenge 1: data visualization Gene Sample 1 … Sample 15 … Sample 22 … A 10 … 9 … 2 … B 0 … 1 … 12 … C 1 … 3 … 2 … Dimensionality of data Gene A Sample 1 Sample 15 Gene C methods from machine learning Sample 22 2D Gene B gene expression level Acute leukemia is one of the rare cancers that develops at early age Cells actively regulate which genes are read: genomics techniques allow monitoring the molecular machinery involved in the process e.g. binding of regulatory proteins: ChIP-seq Recruitment of polymerase -> activation of RNA production: GRO-seq GRO-seq signal represents the level of actively transcribing RNA polymerases at a given genomic position Studying leukemia initiation with regulatory genomics Global run-on sequencing primary transcripts enhancer activity Secondary genetic lesions in ALL • Integration of whole-genome sequencing (WGS) data on ALL structural variation (SV) and Global Run-on (GRO)-sequencing signal Additional genetic lesions in leukemia development SV are frequent TEL-AML1 patients data from Papaemmanuil et al Nat Genet 2014 RAG functions in recombination de Villartay et al. Nat Rev Immun 3, 962-972 2003 Which regions of the genome are susceptible to SV? The basic structural units of chromatin are topologically associated domains (TADs) Topological domains in the mouse ES cell genome (credit: Jesse R. Dixon et al./Nature) Summary 3 Heinäniemi et al. manuscript in preparation Making resources available: TCGA multilevel data & 10K transcriptomes Visualization, data management & software engineering Stuart Kauffman Acknowledgements Ilya Shmulevich Sui Huang Acknowledgements Ville Hautamäki Arto Mannermaa Matti Nykter Anna-Liisa Levonen Ritva Vanninen Minna Kaikkonen Olli Lohi Acknowledgements Krista Tapio Petri Juha Maria