vizome_map.Rmd

---
title: "Vizome dissemination map figure for U24 grant"
author: "Pierrette Lo"
date: "11/7/2019"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)

library(tidyverse)
library(fiftystater)
library(viridis)
```

Data sources:

- U54 DSRN: https://ncihub.org/groups/drsn
- CTD2: https://ocg.cancer.gov/programs/ctd2/centers
- CTSA (CD2H): https://ncats.nih.gov/files/ctsa-funding-information.pdf
  - click on Hub awards for 2019 (60 UL1 awards), export from NIH RePORTER
- NCI-Designated Cancer Centers (CI4CC): https://cancercenters.cancer.gov/Center/CCList
- Look up partner centers on Google Maps -> right click -> What's here? -> c&p lat & long 

Excel cleaning:

- Pulled data and mostly manually inputted states for each center
- Quick manual correction for centers with multiple spellings/wordings of name - C&P one version for all

Read in data
```{r}
data <- readxl::read_xlsx("centers.xlsx")
point_data <- readxl::read_xlsx("partner_coords.xlsx")
```

Clean up center data
```{r}
# make lookup table to convert state abbs to names from {state} dataset
# add dc from {fifty_states} (not contained in {state}) 
state_lookup <- data.frame(state.abb, state = tolower(state.name), stringsAsFactors = F) %>% 
  rbind(c("DC", "district of columbia"))

# add full state names to centers
centers <- data %>% 
  left_join(state_lookup, by = c("state_abb" = "state.abb"))

# lowercase center names for consistency
centers$center <- tolower(centers$center)
```

Version 1: format data for mapping, all states coloured
```{r}
# remove duplicate centers (so plotting no. of unique centers per state)
# get center count per state
# add missing states (center_count will = NA), otherwise there will be holes in the map
mapdata <- centers %>%
  distinct(center, .keep_all = T) %>% 
  group_by(state) %>% 
  summarize(center_count = n()) %>% 
  ungroup() %>% 
  right_join(state_lookup, by = "state") 

# replace NAs with 0 
mapdata$center_count <- replace_na(mapdata$center_count, 0)
```

Version 2: keep states with 0 centers separate so they can be grey
```{r}
mapdata2 <- centers %>%
  distinct(center, .keep_all = T) %>% 
  group_by(state) %>% 
  summarize(center_count = n()) %>% 
  ungroup()

zero_states <- state_lookup %>% 
  anti_join(mapdata2, by = "state")
```

Add state counts to point data
```{r}
point_data <- point_data %>% 
  add_count(state_abb)
```

Version 1 map:
unique centers per state (network + partners) 
plot partner locations on top (size by number since a couple in PDX, seattle, etc. are overlapping)

```{r}
mapdata %>% 
  ggplot() +
  geom_map(aes(map_id = state, fill = center_count), 
           map = fifty_states, color = "white") +
  geom_point(data = point_data, 
             aes(x = long, y = lat, size = n), 
             shape = 21, fill = "darkorange") +
  expand_limits(x = fifty_states$long, y = fifty_states$lat) +
  coord_map() +
  scale_fill_viridis(breaks = c(0, 5, 10, 15), 
                     name = "Network centers\nper state") +
  scale_x_continuous(breaks = NULL) + 
  scale_y_continuous(breaks = NULL) +
  scale_size_continuous(breaks = c(1, 2), 
                        range = c(3,5), 
                        name = "Collaborators\nper state") +
  labs(x = NULL, y = NULL) +
  theme(panel.background = element_blank(),
        plot.caption = element_text(hjust = 0.5, size = 10))

ggsave("vizome_map.png", height = 5, width = 9)
```

Version 2 map: zero_states are grey (easier to see)
```{r}
  ggplot() +
  geom_map(data = mapdata2,
           aes(map_id = state, fill = center_count), 
           map = fifty_states, color = "white") +
  geom_map(data = zero_states,
           aes(map_id = state),
           map = fifty_states, 
           fill = "lightgrey",
           color = "white") +
  geom_point(data = point_data, 
             aes(x = long, y = lat, size = n), 
             shape = 21, fill = "darkorange") +
  expand_limits(x = fifty_states$long, y = fifty_states$lat) +
  coord_map() +
  scale_fill_viridis(breaks = c(1, 7, 13), 
                     name = "Network centers\nper state\n(grey = 0)") +
  scale_x_continuous(breaks = NULL) + 
  scale_y_continuous(breaks = NULL) +
  scale_size_continuous(breaks = c(1, 2), 
                        range = c(3,5), 
                        name = "Collaborators\nper state") +
  labs(x = NULL, y = NULL) +
  theme(panel.background = element_blank(),
        plot.caption = element_text(hjust = 0.5, size = 10))

ggsave("vizome_map_withgrey.png", height = 5, width = 9)
```