clustree: a package for producing clustering trees using ggraph

Clustering analysis is commonly used in many fields to group together similar samples. Many clustering algorithms exist, but all of them require some sort of user input to set parameters that affect the number of clusters produced. Deciding on the correct number of clusters for a given dataset is a difficult problem that can be tackled by looking at the relationships between samples at different resolutions. Here I will present clustree, an R package for producing clustering tree visualisations. These visualisations combine information from multiple clusterings with different resolutions, showing where new clusters come from and how samples change clusters as the number of clusters increases. Summarised information describing the samples in each cluster can be overlaid on the tree to give additional insight. I will also describe my experience developing clustree, particularly how I have made use of the ggraph package. The clustree package is available at https://github.com/lazappi/clustree and a preprint describing clustering trees can be read at https://www.biorxiv.org/content/early/2018/03/02/274035.

This talk was presented at userR! 2018 in Brisbane.

Transcript

1. clustree
   producing clustering trees using ggraph
   Luke Zappia
   @_lazappi_
   lazappi.github.io/clustree
   

   View Slide

2. My data
   OpenStax College, CC BY 3.0 via Wikimedia Commons
   Single-cell RNA-sequencing
   Gene activity in thousands of cells
   ~20000 features (genes)
   ~8000 samples (cells)
   Look for different cell types
   

   View Slide

3. How many
   clusters?
   

   View Slide

4. View Slide

5. Sample K1 K2 K3
   0 A A A
   1 A B C
   2 A A A
   3 A A B
   4 A B A
   5 A A B
   6 A B C
   7 A A A
   8 A A B
   9 A B C
   

   View Slide

6. View Slide

7. View Slide

8. How do we
   do that in R?
   

   View Slide

9. Clusters + transitions
   ID Resolution Cluster Size
   1A 1 A 10
   2A 2 A 6
   2B 2 B 4
   3A 3 A 4
   3B 3 B 3
   3C 3 C 3
   From To Number
   1A 2A 6
   1A 2B 4
   2A 3A 3
   2A 3B 3
   2B 3A 1
   2B 3C 3
   

   View Slide

10. ggplot?
    

    View Slide

11. View Slide

12. Building a graph
    igraph::from_data_frame(edges, vertices = nodes)
    tidygraph::tbl_graph(edges, nodes)
    graph %>%
    activate(nodes) %>%
    filter(...) %>%
    mutate(...) %>%
    activate(edges) %>%
    filter(...) %>%
    mutate(...)
    

    View Slide

13. ggraph
    ggraph(graph, layout = “tree”) +
    geom_edge_link(...) +
    geom_node_point(...) +
    scale_size(...) +
    scale_edge_colour(...)
    

    View Slide

14. clustree(clusterings, ...)
    

    View Slide

15. What does it
    look like?
    

    View Slide

16. Simulations - 1 group
    

    View Slide

17. Simulations - 4 groups
    

    View Slide

18. The Iris dataset
    Tiia Monto CC BY-SA 4.0,
    via Wikimedia Commons
    C T Johansson CC BY 3.0,
    via Wikimedia Commons
    Jefficus,
    via Wikimedia Commons
    Iris setosa
    Iris versicolor
    Iris virginica
    

    View Slide

19. Iris dataset
    k-means clustering
    k = 1,...,5
    

    View Slide

20. View Slide

21. Petal length
    

    View Slide

22. Organoid
    data
    

    View Slide

23. NPHS1
    

    View Slide

24. View Slide

25. t-SNE 2
    t-SNE 1
    t-SNE 1
    t-SNE 2
    

    View Slide

26. Acknowledgements
    Everyone that makes tools and data available
    MCRI Bioinformatics
    Belinda Phipson
    MCRI KDDR
    Alex Combes
    @_lazappi_
    lazappi.github.io/clustree
    install.packages(“clustree”)
    Paper
    doi.org/10.1101/274035
    Slides
    tinyurl.com/clustree-useR2018
    Supervisors
    Alicia Oshlack
    Melissa Little
    

    View Slide