srageo

00_srageo.Rmd

---
title: "Geospatial origins of SRA bioSamples"
output: pdf_notebook
---

# Dependencies

```{r}
library(ggplot2)
library(ggExtra)
library(gridExtra)
library(dplyr)
library(sf)
library(viridisLite)
library(scales)
library(rnaturalearth)
library(rnaturalearthdata)
library(gridExtra)
```

# Download/Load Data (victorlin)

Geo data was scrapped from bioSample XML by vlin.
Repo: https://github.com/serratus-bio/biosample-sql

csv file is hosted on S3

```{r}
# Download dataset
library(utils)
download.file(url = 'https://serratus-public.s3.amazonaws.com/geo/vl_geo_210224.csv.gz',
              destfile = './vl_geo_210224.csv.gz',
              method = 'wget' )

```


```{r}
# Load parsed lon-lat data (VL)
coord.df <- read.table(file = 'vl_geo_210224.csv.gz',
                      sep = ',', header = T)
  colnames(coord.df) <- c('SRA', 'BioSample', 'date', 'lon', 'lat', 'geotext')
  
# Retain only lat/lon if it's within worldmap coordinates
  coord.df <- coord.df[ (coord.df$lat > -90 & coord.df$lat < 90), ]
  coord.df <- coord.df[ (coord.df$lon > -180 & coord.df$lon < 180), ]
  
 ```
 
 # Histogram of lat/lon
 
 ```{r} 
# VN SQL data
lon.hist <- ggplot( coord.df, aes(lon)) + 
  geom_histogram(bins = 360) +
  xlim(c(-180,180)) +
  theme_bw()

lat.hist <- ggplot( coord.df, aes(lat)) + 
  geom_histogram(bins = 360) +
  xlim(c(-90,90)) + coord_flip() + 
  theme_bw()

grid.arrange(lat.hist, lon.hist)

```

# Europe Heatmap

```{r}
# Plot Europe -- DARK
world <- ne_countries(scale = "medium", returnclass = "sf")

# Hex with viridis
euro  <- ggplot(data = world) +
  geom_sf(fill = 'black', color = 'gray5') +
  theme(panel.background = element_rect( fill = 'gray10'),
        panel.grid.major = element_line(color = 'gray20',
                                        linetype = 'dashed',
                                        size = 0.2)) +
  geom_hex( bins = 1800, data = coord.df,
            aes(x = lon, y = lat)) +
  coord_sf(xlim = c(-15,25), ylim = c(30,60)) +
  scale_fill_continuous(type = "viridis", option = "plasma", trans = "log2")
euro
```

# Worldmap heatmap

```{r}
# Plot Worldmap -- DARK
world <- ne_countries(scale = "medium", returnclass = "sf")

# Hex with viridis
earth <- ggplot(data = world) +
  geom_sf(fill = 'black', color = 'gray5') +
  theme(panel.background = element_rect( fill = 'gray10'),
        panel.grid.major = element_line(color = 'gray20',
                                        linetype = 'dashed',
                                        size = 0.2)) +
  geom_hex( bins = 720, data = coord.df,
            aes(x = lon, y = lat)) +
  scale_fill_continuous(type = "viridis", option = "plasma", trans = "log2")
earth
```

01_europe.png

02_planet.png