Tools and techniques for single-cell RNA sequencing data

RNA sequencing of individual cells allows us to take a snapshot of the dynamic processes within a cell and explore the differences between cell types. As this technology has developed over the last few years it has been rapidly adopted by researchers in areas such as developmental biology. Along with the development of protocols for producing this data has been a simultaneous burst in the development of computational methods for analysing it. My thesis explores the computational tools and techniques for analysing single-cell RNA-sequencing data. I will present a database that charts the release of analysis software (https://scrna-tools.org), Splatter, a software package for easily simulating single-cell datasets from multiple models (http://bioconductor.org/packages/splatter/) and clustering trees, a visualisation approach for inspecting clustering results at multiple resolutions (https://CRAN.R-project.org/package=clustree). In the final part of my thesis, I use an analysis of a kidney organoid dataset to demonstrate and compare some of the current analysis methods.

Transcript

1. TOOLS AND TECHNIQUES
   FOR SINGLE-CELL
   RNA SEQUENCING DATA
   LUKE ZAPPIA
   

   View Slide

2. 1 Introduction
   2 Tools
   3 Simulations
   4 Clustering trees
   5 Analysis
   6 Conclusion
   1
   2
   3
   4
   5
   6
   

   View Slide

3. Introduction
   Matthew Daniels via The Cell Image Library
   http://www.cellimagelibrary.org/images/38912
   1
   

   View Slide

4. View Slide

5. ACTGACTCCA
   TCAGTACTGA
   CGTGTCATAG
   GATTGACCTA
   Gene Sample 1
   A 43
   B 3
   C 17
   D 24
   RNA sequencing
   

   View Slide

6. Alignment-free
   quantification
   Alignment
   Counting
   Aligned reads
   Raw reads
   Reference
   genome
   Gene
   annotation
   Reference
   transcriptome
   Expression matrix
   Normalisation
   Differential expression testing
   Gene set testing
   Visualisation
   Gene sets
   Interpretation
   Quality control
   

   View Slide

7. Svensson et al. DOI: 10.1038/nprot.2017.149
   

   View Slide

8. mccarrolllab.com/dropseq/
   Droplet cell capture
   

   View Slide

9. scRNA-seq
   ACTGACTCCA
   TCAGTACTGA
   CGTGTCATAG
   GATTGACCTA
   ACTGACTCCA
   TCAGTACTGA
   CGTGTCATAG
   GATTGACCTA
   ACTGACTCCA
   TCAGTACTGA
   CGTGTCATAG
   GATTGACCTA
   ACTGACTCCA
   TCAGTACTGA
   CGTGTCATAG
   GATTGACCTA
   Gene Cell 1 Cell 2 Cell 3 Cell 4
   A 12 10 9 0
   B 0 0 0 1
   C 9 6 0 0
   D 7 0 4 0
   

   View Slide

10. View Slide

11. View Slide

12. Kidney development
    Images from OpenStax College, CC BY 3.0 via Wikimedia Commons
    

    View Slide

13. Kidney organoids
    Day 0 4
    7 10 18 25
    CHIR FGF9
    FGF9
    CHIR
    Form pellets
    No GF
    iPSCs organoid
    

    View Slide

14. Aims
    1. Understand the computational tools used to
    analyse scRNA-seq data
    2. Contribute to tool development
    3. Apply tools to a kidney organoid dataset
    

    View Slide

15. Tools
    Andres J Garcia and Ankur Singh via The Cell Image Library
    2
    http://www.cellimagelibrary.org/images/44701
    

    View Slide

16. View Slide

17. www. .org
    

    View Slide

18. View Slide

19. Number of tools Publication status
    

    View Slide

20. Software licenses Platforms
    

    View Slide

21. Analysis categories
    

    View Slide

22. Users over time
    By country Top 10 By continent
    

    View Slide

23. “Exploring the single-cell RNA-seq analysis landscape with the scRNA-tools database”
    PLoS Computational Biology (2018) DOI: 10.1371/journal.pcbi.1006245
    

    View Slide

24. Simulations
    3
    S. Schuller via The Cell Image Library
    http://www.cellimagelibrary.org/images/38903
    

    View Slide

25. Simulations
    Provide a truth to test against
    But
    - Often poorly documented and explained
    - Not easily reproducible or reusable
    - Don’t demonstrate similarity to real data
    

    View Slide

26. Bioconductor package
    Consistent, easy-to-use interface
    Multiple simulation models
    

    View Slide

27. Splat model
    Negative binomial
    Expression outliers
    Defined library sizes
    Mean-variance trend
    Dropout
    

    View Slide

28. Other models
    Simple - Negative binomial
    Lun - NB with cell factors
    DOI: 10.1186/s13059-016-0947-7
    Lun2 - Sampled NB with batch effects
    DOI: 10.1093/biostatistics/kxw055
    scDD - NB with bimodality
    DOI: 10.1186/s13059-016-1077-y
    BASiCS - NB with spike-ins
    DOI: 10.1371/journal.pcbi.1004333
    mfa - Bifurcating pseudotime trajectory
    DOI: 10.12688/wellcomeopenres.11087.1
    PhenoPath - Pseudotime with gene types
    DOI: 10.1038/s41467-018-04696-6
    ZINB-WaVE - Sophisticated ZINB
    DOI: 10.1186/s13059-018-1406-4
    SparseDC - Clusters across two conditions
    DOI: 10.1093/nar/gkx1113
    

    View Slide

29. Real data Parameters Dataset
    Estimation Simulation
    params1 <- splatEstimate(real.data)
    params2 <- simpleEstimate(real.data)
    sim1 <- splatSimulate(params1, ...)
    sim2 <- simpleSimulate(params2, ...)
    datasets <- list(Real = real.data,
    Splat = sim1,
    Simple = sim2)
    comp <- compareSCESets(datasets)
    diff <- diffSCESets(datasets, ref = “Real”)
    1. Estimate
    2. Simulate
    3. Compare
    

    View Slide

30. ZINB-WaVE
    SparseDC
    PhenoPath
    mfa
    BASiCS
    scDD
    Lun2 (ZINB)
    Lun2
    Lun
    Simple
    Splat (Drop)
    Splat
    Real
    Mean log
    2
    (CPM + 1)
    Distribution of mean expression
    ZINB-WaVE
    SparseDC
    PhenoPath
    mfa
    BASiCS
    scDD
    Lun2 (ZINB)
    Lun2
    Lun
    Simple
    Splat (Drop)
    Splat
    Rank Difference Mean log
    2
    (CPM + 1)
    Difference in mean expression
    

    View Slide

31. ZINB-WaVE
    SparseDC
    PhenoPath
    mfa
    BASiCS
    scDD
    Lun2 (ZINB)
    Lun2
    Lun
    Simple
    Splat (Drop)
    Splat
    Mean
    Variance
    Mean-Variance
    Library size
    % Zeros (Cell)
    % Zeros (Gene)
    Mean-Zeros
    Rank of MAD from real data
    

    View Slide

32. Complex simulations
    Groups Batches Paths
    

    View Slide

33. “Splatter: simulation of single-cell RNA sequencing data”
    Genome Biology (2017) DOI: 10.1186/s13059-017-1305-0
    

    View Slide

34. Clustering trees
    4
    http://www.cellimagelibrary.org/images/40483
    M Uhlen et al. via The Cell Image Library
    

    View Slide

35. How many clusters?
    Low resolution
    (fewer clusters)
    High resolution
    (more clusters)
    

    View Slide

36. A tree of clusters?
    

    View Slide

37. View Slide

38. View Slide

39. Weighting edges
    In proportion =
    Number of cells
    on edge
    Number of cells in
    higher res cluster
    

    View Slide

40. View Slide

41. View Slide

42. Gene expression
    

    View Slide

43. Cell cycle SC3 stability
    Number of
    genes
    

    View Slide

44. View Slide

45. t-SNE 2
    t-SNE 1
    t-SNE 1
    t-SNE 2
    

    View Slide

46. “Clustering trees: a visualisation for evaluating clusterings at multiple resolutions”
    GigaScience (2018) DOI: gigascience/giy083
    

    View Slide

47. Analysis
    Natalie Prigozhina via The Cell Image Library
    http://www.cellimagelibrary.org/images/48101
    5
    

    View Slide

48. GATA3
    ECAD
    LTL
    WT1
    CD +
    DT +
    PT +
    Glo
    

    View Slide

49. Dataset
    4 Organoids
    10x Chromium
    2 Batches (3 + 1)
    7937 cells (6649 + 1288)
    Identify cell types
    Alignment
    Quantification
    Quality control
    Integration
    Clustering
    Gene detection
    CellRanger
    CellRanger
    scater
    Seurat
    Seurat
    Seurat
    Analysis steps
    

    View Slide

50. Stroma Endothelium
    Cell cycle Podocyte
    Epithelium
    

    View Slide

51. Glial
    Neural progenitor
    Muscle
    progenitor
    

    View Slide

52. Human dataset
    16 week fetal kidney
    3178 cells
    10x Chromium
    Lindström et al. “Conserved and Divergent Features of Mesenchymal Progenitor
    Cell Types within the Cortical Nephrogenic Niche of the Human and Mouse Kidney”
    J Am Soc Nephrol (2018) DOI:10.1681/ASN.2017080890
    

    View Slide

53. Stroma
    Fetal kidney
    Stroma
    Organoid
    Stroma
    Stroma
    Endothelium
    Cell cycle
    Nephron progenitor
    Podocyte
    Cell cycle
    Stroma
    Nephron
    Glial
    Immune
    Blood
    Neural progenitor
    Podocyte
    1000
    750
    500
    250
    0
    1250
    1000
    750
    500
    250
    0
    1250
    Number of cells
    

    View Slide

54. “Single-cell analysis reveals congruence between kidney organoids and human fetal kidney”
    Genome Medicine (2019) DOI: 10.1186/s13073-019-0615-0
    

    View Slide

55. What if we did
    things
    differently?
    

    View Slide

56. Droplet selection
    ~ 1 million droplets
    

    View Slide

57. View Slide

58. Quality control
    Manual thresholds PCA based
    

    View Slide

59. Gene selection
    Seurat
    

    View Slide

60. Gene selection
    M3Drop
    

    View Slide

61. Gene selection
    Overlap
    Seurat only
    M3Drop only
    Both
    

    View Slide

62. Comparison
    

    View Slide

63. Marker genes
    

    View Slide

64. Partition-based graph abstraction
    Cell graph PAGA cluster graph
    

    View Slide

65. Cell velocity
    AAAAAAA
    Unspliced RNA
    Mature mRNA
    

    View Slide

66. Podocyte
    Epithelial
    Endothelial
    Stroma
    Neural
    Muscle
    Immune?
    

    View Slide

67. Summary
    New droplet selection methods can give many more cells
    Seurat clustering is robust to gene selection
    Possible immune-like population
    Alternative methods can help interpretation
    

    View Slide

68. Natalie Prigozhina via The Cell Image Library
    http://www.cellimagelibrary.org/images/48108
    Conclusion
    6
    

    View Slide

69. What did I do?
    Build a database of scRNA-seq analysis tools and a website to interact
    with it
    Develop a software package for simulating scRNA-seq data and a
    flexible simulation model
    Design an algorithm for visualising clustering at multiple resolutions
    and a software package that implements it
    Perform an analysis of kidney organoid data to profile the cell types
    present and demonstrate the effect of different tools and decisions
    

    View Slide

70. What next?
    Bigger datasets, more computation
    Convergence on methods that work
    Continued development of software
    Reference datasets
    Integration of data types
    Spatial transcriptomics
    

    View Slide

71. Acknowledgments
    Supervisors
    Alicia Oshlack
    Melissa Little
    Committee
    Andrew Pask
    Christine Wells
    Edmund Crampin
    Everyone that makes their tools and data available
    MCRI Bioinformatics
    Belinda Phipson
    Breon Schmidt
    MCRI KDDR
    Alex Combes
    COMBINE
    Friends and family
    Developers
    dplyr, ggplot2, scater,
    scran, Seurat, workflowr,
    rmarkdown, knitr,
    tidygraph, ggraph, edgeR...
    

    View Slide

72. install.packages(“clustree”)
    Paper: 10.1093/gigascience/giy083
    Paper: doi.org/10.1186/s13059-017-1305-0
    biocLite(“splatter”)
    Paper: doi.org/10.1093/gigascience/giy083
    www.scRNA-tools.org
    oshlacklab.com/combes-organoid-paper
    Paper: doi.org/10.1186/s13073-019-0615-0
    @_lazappi_
    oshlacklab.com
    github.com/lazappi
    

    View Slide