An open data exchange for cell migration research
- 1. CC BY-SA 4.0 TOWARDS AN OPEN DATA EXCHANGE ECOSYSTEM: FORGING A NEW PATH FOR CELL MIGRATION DATA ANALYSIS AND MINING 18 October 2016 public PhD defense - paola masuzzo
- 4. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 5. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 6. CC BY-SA 4.0 Cell migration is necessary for many physiological functions Basement membrane Wound Migrating epithelial cells
- 7. CC BY-SA 4.0 Cell migration is necessary for many physiological functions Blood vessel Site of tissue injury Migrating neutrophil Basement membrane Wound Migrating epithelial cells
- 8. CC BY-SA 4.0 Unfortunately, it is also implicated in many diseases, such as metastatic cancer
- 9. CC BY-SA 4.0 The cytoskeleton is the key structural framework responsible for cell migration Adapted from Herzog et al., Cell Bio Lab Handbook, 1994
- 10. CC BY-SA 4.0 The cytoskeleton is the key structural framework responsible for cell migration Adapted from Herzog et al., Cell Bio Lab Handbook, 1994
- 11. CC BY-SA 4.0 Different actin filament structures are essential for cell migration Actin network Filopodia Lamellipodium movement Leading edge 2D migration
- 12. CC BY-SA 4.0 Different actin filament structures are essential for cell migration Actin network Filopodia Lamellipodium movement Leading edge 2D migration 3D invasion Basement membrane Epithelial cell Tumor cell Extracellular matrix Invadopodia Protruding bleb Lamellipodia or pseudopodia
- 13. CC BY-SA 4.0 Cell translocation depends on a cyclic interplay between cell adhesion and de-adhesion
- 14. CC BY-SA 4.0 A typical experimental workflow for a cell migration study is composed of diverse steps Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0 sample preparation image acquisition image processing data analysis
- 15. CC BY-SA 4.0 A typical experimental workflow for a cell migration study is composed of diverse steps Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0 sample preparation image acquisition image processing data analysis
- 16. CC BY-SA 4.0 Several assays are available for in vitro assessment of cell migration Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0 sample preparation image acquisition image processing data analysis Wound-healing assay pipette tip scratch Cell-exclusion zone assay cell-free zone silicone stopper Spheroid assay multicellular spheroid
- 17. CC BY-SA 4.0 Live-cell phase-contrast and fluorescence microscopy generates quantifiable data Servier Medical Art, CC-BY 3.0; Cell Image Library, Public Domain; Mierke et al., 2011 sample preparation image acquisition image processing data analysis
- 18. CC BY-SA 4.0 Time-lapse video microscopy captures cell migration dynamics time
- 19. CC BY-SA 4.0 Image processing is a multi-step operation comprising segmentation and tracking Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0, Harder et al., 2015 image pre-processing cell tracking cell segmentation sample preparation image acquisition image processing data analysis
- 20. CC BY-SA 4.0 Quantitative parameters are then extracted for cell sheet and single-cell trajectories Image processed with CELLMIA, UGent (Van Troys M, Ampe C) and DciLabs area in time µm²/min coordinates (x, y, t) µm/min
- 21. CC BY-SA 4.0 Ultimately, data analysis enables interpretation of the experiment Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0; O’ Brien et al., 2014 sample preparation image acquisition image processing data analysis
- 22. CC BY-SA 4.0 The Ghent platform enables automation of high-throughput cell migration experiments Phase-contrast live-cell imaging time-lapse: 16-48 h interval: 15-20 min Adapted from Lynn Huyck, PhD thesis, 2012 (promoter Van Troys M)
- 23. CC BY-SA 4.0 Images are automatically processed Adapted from Lynn Huyck, PhD thesis, 2012 (promoter Van Troys M); images processed with CELLMIA, UGent and DciLabs t = 0h t = 24h t = 0h t = 36h
- 24. CC BY-SA 4.0 Such high-throughput experiments produce complex and rich data sets Servier Medical Art, CC-BY 3.0; Cell Image Library, CC-BY 3.0 sample preparation image acquisition image processing data analysis • paper laboratory notebooks • electronic laboratory notebooks • spreadsheets • text files • protocols • papers... • raw files • XML files • proprietary microscope or acquisition software files ND2 for Nikon, LIF for Leica, OIB or OIF for Olympus, LSM or ZVI for Zeiss • image files with pixel values and metadata • png, jpeg, tiff, avi • text files describing processing algorithms • text files describing extracted features • graphs, plots • analysis pipelines • text files describing computational algorithms...
- 25. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 26. CC BY-SA 4.0 The overall objective of this PhD is to advance bioinformatics for cell migration Cell migration experiments have become de facto high-throughput, but bioinformatics has lagged behind Due to lack of automated systems and appropriate algorithms, a big proportion of cell migration data is still not exploited The heterogeneity of the field hampers open data exchange, impeding advanced data analysis and mining
- 27. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 28. CC BY-SA 4.0 CellMissy is our open-source tool for cell migration data management and analysis 0 3h 6h wound cells Experiment Data Analyzer Data Loader Collective cell migration Single-cell migration Experiment Manager Masuzzo et al., Bioinformatics, 2013; https://github.com/compomics/cellmissy
- 29. CC BY-SA 4.0 CellMissy guides and captures the experimental setup
- 30. CC BY-SA 4.0 This experimental setup encloses detailed metadata annotation
- 31. CC BY-SA 4.0 CellMissy can automatically import all the data and metadata
- 32. CC BY-SA 4.0 All the data are then stored in a structured way in a relational database
- 33. CC BY-SA 4.0 CellMissy enables efficient data exploration and analysis time (min) Area(μm2 ) wound area cell-covered area
- 34. CC BY-SA 4.0 A primary focus of data analysis is statistical comparison of samples analysis report with graphs, tables and results cell sheet velocity (µm²/min)
- 35. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 36. CC BY-SA 4.0 Cell migration can occur in both collective and individual fashion Collective migration Multicellular streaming Mesenchymal Amoeboid (blebs) Amoeboid (pseudopodia, filopodia) INDIVIDUAL MIGRATION COLLECTIVE MIGRATION Adapted from Friedl et al., J. Exp. Med., 2010
- 37. CC BY-SA 4.0 Many informative parameters can be derived from single-cell trajectories Masuzzo et al., under review, 2016 x y single cell
- 38. CC BY-SA 4.0 Many informative parameters can be derived from single-cell trajectories Masuzzo et al., under review, 2016 Euclidean distance Cumulative distance x y single cell
- 39. CC BY-SA 4.0 Many informative parameters can be derived from single-cell trajectories Masuzzo et al., under review, 2016 Euclidean distance Cumulative distance x y single cell parameter mathematical description di: instantaneous displacement of the cell centroid between adjacent time points 𝑑𝑖 = 𝑥𝑖+1 − 𝑥𝑖 2 + 𝑦𝑖+1 − 𝑦𝑖 2 si: instantaneous speed between adjacent time points 𝑠𝑖 = 𝑑(𝑝𝑖, 𝑝𝑖+1) ∆𝑡 αi: turning angle between consecutive steps 𝛼𝑖 = 𝑡𝑎𝑛−1 𝑦𝑖+1 − 𝑦𝑖 𝑥𝑖+1 − 𝑥𝑖 dtot: cumulative distance, total distance travelled 𝑑 𝑡𝑜𝑡 = 𝑖=1 𝑁−1 𝑑 𝑝𝑖, 𝑝𝑖+1 dnet: Euclidean distance, net distance travelled 𝑑 𝑛𝑒𝑡 = 𝑑 𝑝1, 𝑝 𝑁 ep_dr: end-point directionality ratio (confinement ratio, meandering index) 𝑒𝑝_𝑑𝑟 = 𝑑 𝑛𝑒𝑡 𝑑 𝑡𝑜𝑡 MD: median displacement 𝑀𝐷 = 𝑚𝑒𝑑𝑖𝑎𝑛 𝑑𝑖 MS: median speed 𝑀𝑆 = 𝑚𝑒𝑑𝑖𝑎𝑛 𝑠𝑖 MTA: median turning angle 𝑀𝑇𝐴 = 𝑚𝑒𝑑𝑖𝑎𝑛 𝛼𝑖
- 40. CC BY-SA 4.0 Step-centric parameters are aggregated values of all migration steps
- 41. CC BY-SA 4.0 Trajectory-centric parameters are instead computed on each cell and then averaged for the cell population ... trajectory 1 trajectory 2 trajectory 3 trajectory 4 trajectory 5
- 42. CC BY-SA 4.0 The new single-cell module allows for both these computations to take place Masuzzo et al., under review, 2016 ... trajectory 1 trajectory 2 trajectory 3 trajectory 4 trajectory 5 trajectory-centric parameters trajectory displacement (µm) density step displacement (µm) density pool of migration steps step-centric parameters
- 43. CC BY-SA 4.0 A flexible two-step filtering criterion is implemented for data quality control Masuzzo et al., under review, 2016
- 44. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 45. CC BY-SA 4.0 More advanced features are needed to describe the complexity of the phenomenon Masuzzo et al., in preparation, 2016
- 46. CC BY-SA 4.0 The enclosing circle set is a new way to describe local structure of trajectories Masuzzo et al., in preparation, 2016 radius: 6 µm nr_circles: 14 direction of motion
- 47. CC BY-SA 4.0 The enclosing circle set is a new way to describe local structure of trajectories Masuzzo et al., in preparation, 2016 radius: 6 µm nr_circles: 14 direction of motion radius: 3 µm nr_circles: 22 direction of motion
- 48. CC BY-SA 4.0 The fractal dimension is derived from the enclosing circle set 𝐹𝐷 𝑆 = lim 𝑟→0 log )𝑁(𝑟 log 1 𝑟 Masuzzo et al., in preparation, 2016 FD=0.35 FD=0.83
- 49. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 50. CC BY-SA 4.0 Data and metadata exchange options are already available in CellMissy lab A
- 51. CC BY-SA 4.0 Data and metadata exchange options are already available in CellMissy lab A lab B
- 52. CC BY-SA 4.0 Data and metadata exchange options are already available in CellMissy lab B This is one file in CellMissy! (≈10 MB) lab A
- 53. CC BY-SA 4.0 But we can easily extend this concept to a bigger scale Data Repository Local Software
- 54. CC BY-SA 4.0 The seed of this idea was planted in the field Friedl et al., Nature Reviews, 2012
- 55. CC BY-SA 4.0 It only needed some water to grow Cell migration workshop, Ghent, March 2014; Masuzzo et al., Trends in Cell Biology, 2015
- 56. CC BY-SA 4.0 An open data exchange ecosystem for cell migration research is now on its way Masuzzo et al., Trends in Cell Biology, 2015
- 57. CC BY-SA 4.0 An open data exchange ecosystem for cell migration research is now on its way Masuzzo et al., Trends in Cell Biology, 2015
- 58. CC BY-SA 4.0 An open data exchange ecosystem for cell migration research is now on its way Masuzzo et al., Trends in Cell Biology, 2015
- 59. CC BY-SA 4.0 This open data ecosystem falls into the broader context of open science Knoth and Pontika, Open Science Taxonomy, figshare, 2015
- 60. CC BY-SA 4.0 Open access is another key factor in the open science equation Knoth and Pontika, Open Science Taxonomy, figshare, 2015
- 61. CC BY-SA 4.0 The impacts of open access are very broad and affect many areas Tennant, Masuzzo et al., F1000Research, 2016; Wikimedia Commons, Public Domain
- 62. CC BY-SA 4.0 Publish your work open access can bring you enormous benefits CC-BY Danny Kingsley & Sarah Brown
- 63. CC BY-SA 4.0 Publish your work open access can bring you enormous benefits CC-BY Danny Kingsley & Sarah Brown
- 64. CC BY-SA 4.0 Publish your work open access can bring you enormous benefits CC-BY Danny Kingsley & Sarah Brown
- 65. CC BY-SA 4.0 Open access also allows automatic knowledge extraction through text mining automatically detect a set of core information reported when describing cell migration experiments check for nomenclature consistency, the use of common terms or ontologies to describe the same concept construct a knowledge map to describe the state-of-the-art, especially in terms of cell motility-related compounds and cancer cell lines
- 66. CC BY-SA 4.0 Introduction to cell migration Research problems Results CellMissy: an automated tool for cell migration A new CellMissy module for single-cell analysis Engineering features to describe stochasticity Towards an open data exchange ecosystem Conclusions and future perspectives
- 67. CC BY-SA 4.0 This PhD has tackled key bioinformatics challenges in cell migration research CellMissy is the first free and open-source tool for the management, annotation and storage of cell migration experiments A new dedicated module, together with novel features, enable detailed and more complex quantification of single-cell migration experiments International research efforts are currently spent towards the establishment of an open data exchange ecosystem, opening the way to more advanced data analysis and mining strategies
- 68. CC BY-SA 4.0 These results have paved the way to even more exciting opportunities CellMissy has already been extended with dose-response analysis capabilities, and more development is planned to allow meta-analyses to take place Further engineering, validation, and selection of single-cell migration features is planned; these features will then be used to automatically detect and classify migratory phenotypes Joined efforts of MULTIMOT and the CMSO will ultimately enable global data dissemination in the field, allowing data re-use, re-discovery and re-purpose
- 69. CC BY-SA 4.0
Editor's Notes
- #4: vehicle detection systems that are able to determine the position, size and speed of multiple vehicles in highways - avoid collisions, surveillance systems and so on
- #10: How does this work? The cell has an internal machinery capable make it move. The major component of this machinery is the cytoskeleton. Three filament structure...
- #11: How does this work? The cell has an internal machinery capable make it move. The major component of this machinery is the cytoskeleton. Three filament structure...
- #12: Motility is initiated by an actin-dependent protrusion at the leading edge filopodia (1D finger-like protrusions) and lamellipodia (2D sheet-like protrusions).
- #13: Motility is initiated by an actin-dependent protrusion at the leading edge filopodia (1D finger-like protrusions) and lamellipodia (2D sheet-like protrusions).
- #14: substratum: solid surface to which a cell can adhere; cell needs spatial and functional asymmetry actin polymerization leading edge (front) + trailing edge (rear) 1 - protrusion at the leading edge 2 - new adhesion to the substratum 3 - actomyosin contractility - detachment of contact points cell body translocation 4 - retraction at the trailing edge
- #22: fibrinogen, collagen, fibronectin
- #23: Live-cell imaging Single-cell + cell bulk
- #24: Live-cell imaging Single-cell + cell bulk
- #34: 2D conditions are together, 3D conditions much slower
- #35: statistical comparison of the median cell sheet velocity across biological conditions: from control + drug solvent to different concentrations of latrunculin (important for dose response analysis) latrunculin -- from 65.2 µM to 750.0 µM latrunculin -- disruption of the actin filaments of the cytoskeleton Inhibition of actin polymerization by latrunculin A disrupts actin filament formation, cytoskeletal organization, cell migration, and chemotaxis15,38.
- #37: collective: building and remodeling tissues, cancer invasion, wound healing single: immune processes, but also cancer invasion
- #38: Representation in 2D Euclidean space of single-cell trajectory -- define the parameters in the table. In particular, the turning angle between two points, the ED, and the CD.
- #39: Representation in 2D Euclidean space of single-cell trajectory -- define the parameters in the table. In particular, the turning angle between two points, the ED, and the CD.
- #40: Representation in 2D Euclidean space of single-cell trajectory -- define the parameters in the table. In particular, the turning angle between two points, the ED, and the CD.
- #45: this type of characterization is however not robust enough for more complex and heterogeneous experiments
- #46: minimum bounding box + convex hull - and derived parameters
- #47: how many circles of radius r do we need to cover the set of points?
- #48: how many circles of radius r do we need to cover the set of points?
- #66: contentmine fellowship create new knowledge by combining individual findings