vignettes/pkgdown_only/sankey.md

In epijim/PharmaPlots: Plotting wrappers

Using RochePlot::sankey

Get data

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

    library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

    library(RochePlot)
    library(RocheTeradata)

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

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

## A teradataR connection is now live for blackj9 with no home datalab @rochetd.kau.roche.com.

## I gave you a teradataR style connection, the connection file is called tdConnection.

    d_LoT <- RocheTeradata::tdRWDSquery(
      "
      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
  )

## Please manually validate the numbers - this function is unvalidated.

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
  )

## Please manually validate the numbers - this function is unvalidated.

 str(d_plot)

## List of 2
##  $ edges:'data.frame':   49 obs. of  5 variables:
##   ..$ source      : chr [1:49] "Bendamustine,Rituximab 1L" "Bendamustine,Rituximab 1L" "Bendamustine,Rituximab 1L" "Bendamustine,Rituximab 1L" ...
##   ..$ target      : chr [1:49] "Bendamustine,Rituximab 2L" "Ibrutinib 2L" "No recorded 2L" "Other Tx 2L" ...
##   ..$ value       : int [1:49] 39 142 805 165 16 11 52 168 71 6 ...
##   ..$ idnum_source: num [1:49] 0 0 0 0 0 1 1 1 1 1 ...
##   ..$ idnum_target: num [1:49] 1 4 8 11 14 2 5 9 12 15 ...
##  $ nodes:'data.frame':   16 obs. of  4 variables:
##   ..$ id_char  : chr [1:16] "Bendamustine,Rituximab 1L" "Bendamustine,Rituximab 2L" "Bendamustine,Rituximab 3L" "Ibrutinib 1L" ...
##   ..$ id_factor: Factor w/ 16 levels "Bendamustine,Rituximab 1L",..: 1 2 3 4 5 6 7 8 9 10 ...
##   ..$ id_num   : num [1:16] 0 1 2 3 4 5 6 7 8 9 ...
##   ..$ colour   : chr [1:16] "rgb(30,144,255)" "rgb(30,144,255)" "rgb(30,144,255)" "rgb(30,144,255)" ...

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'"
  )

| id_char | id_factor | id_num| colour | |:--------------------------|:--------------------------|--------:|:----------------| | Bendamustine,Rituximab 1L | Bendamustine,Rituximab 1L | 0| rgb(30,144,255) | | Bendamustine,Rituximab 2L | Bendamustine,Rituximab 2L | 1| rgb(30,144,255) | | Bendamustine,Rituximab 3L | Bendamustine,Rituximab 3L | 2| rgb(30,144,255) | | Ibrutinib 1L | Ibrutinib 1L | 3| rgb(30,144,255) | | Ibrutinib 2L | Ibrutinib 2L | 4| rgb(30,144,255) |

edges

Second is the edges.

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

| source | target | value| idnum_source| idnum_target| |:--------------------------|:--------------------------|------:|--------------:|--------------:| | Bendamustine,Rituximab 1L | Bendamustine,Rituximab 2L | 39| 0| 1| | Bendamustine,Rituximab 1L | Ibrutinib 2L | 142| 0| 4| | Bendamustine,Rituximab 1L | No recorded 2L | 805| 0| 8| | Bendamustine,Rituximab 1L | Other Tx 2L | 165| 0| 11| | Bendamustine,Rituximab 1L | Rituximab 2L | 16| 0| 14|

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
  )

epijim/PharmaPlots documentation built on Nov. 4, 2019, 11:54 a.m.