In epijim/PharmaPlots: Plotting wrappers

Get data

For this example, I will use line data from Flatiron.

    library(dplyr)
    library(RochePlot)
    library(RocheTeradata)


    # data valid today
    valid_date <- Sys.Date()

    # connect to data
    connect_teradata(
      datalab = FALSE,
      pwd = readLines("~/.ssh/.pass")
    )

    d_LoT <- RocheTeradata::query_teradata(
      "
      SELECT
      patientid
      ,LineName
      ,LineNumber
      ,IsMaintenanceTherapy
      FROM rwd_vdm_flatiron.v_cll_lineoftherapy
      WHERE
      VALID_START <= CAST('",paste0(valid_date," 00:00:00"),"' AS TIMESTAMP) AND
      VALID_END >= CAST('",paste0(valid_date," 00:00:00"),"' AS TIMESTAMP)
      ;"
    ) %>%
      filter(
        # just do lines 0 to 3
        LineNumber < 4 &
          # and ignore maintenence
          IsMaintenanceTherapy == "False"
      )

Using plotly

By default, the plot will use plotly to make the Sankey. E.g., the code below.

  sankey(
    dataframe = d_LoT,
    id = "PatientID",
    linename = "LineName",
    linenumber = "LineNumber",
    n_common = 3
  )

Look at the data

I can also just get the function to format the data, put not make the plot.

Here I save that formatted data into d_plot.

  d_plot <- sankey(
    dataframe = d_LoT,
    id = "PatientID",
    linename = "LineName",
    linenumber = "LineNumber",
    n_common = 3,
    return_data = TRUE
  )

 str(d_plot)

nodes

There are two dataframes within the list returned. The first is the nodes.

  knitr::kable(
    d_plot$nodes %>% head(5),
    caption = "First 5 rows of 'nodes'"
  )

edges

Second is the edges.

  knitr::kable(
    d_plot$edges %>% head(5),
    caption = "First 5 rows of 'edges'"
  )

Different sankey wrapper

Using the data returned - I can also try plotting the data using a different plotting function. E.g. here I use a more basic plotting function which is a simpler mapping to Bostock's D3.js library than the default plotly mapping.

  networkD3::sankeyNetwork(
    Links = d_plot$edges,
    Nodes = d_plot$nodes,
    Source = "idnum_source",
    Target = "idnum_target",
    Value = "value",
    NodeID = "id_char",
    units = "people", fontSize = 12, nodeWidth = 30
  )