Some updates to the format (including SRA removal)

.gitignore

@@ -26,3 +26,5 @@ Intermediate/Formatted/GenbankFormatted.csv
 Intermediate/Formatted/GenbankFormatted.csv
 Intermediate/Formatted/VIRIONUnprocessed.csv
 Intermediate/Formatted/GenbankFormatted.csv
+wc5/alt.bil
+wc5/alt.hdr

Code/03_Merge clean files.R

@@ -5,13 +5,14 @@ library(tidyverse); library(magrittr); library(vroom)
 
 gb <- vroom("Intermediate/Formatted/GenbankFormatted.csv.gz")
 clo <- read_csv("Intermediate/Formatted/CloverFormatted.csv")
-sra <- read_csv("Intermediate/Formatted/SRAFormatted.csv")
+#sra <- read_csv("Intermediate/Formatted/SRAFormatted.csv")
 pred <- read_csv("Intermediate/Formatted/PredictFormatted.csv")
 globi <- read_csv("Intermediate/Formatted/GLOBIFormatted.csv")
 
 if(class(clo$NCBIAccession)=='numeric') {clo %<>% mutate(NCBIAccession = as.character(NCBIAccession))}
 
-virion <- bind_rows(clo, pred, gb, sra, globi)
+# virion <- bind_rows(clo, pred, gb, sra, globi)
+virion <- bind_rows(clo, pred, gb, globi)
 
 vroom_write(virion, "Intermediate/Formatted/VIRIONUnprocessed.csv.gz")
 

Code/04_High level VIRION checks.R

@@ -62,8 +62,18 @@ virion %<>%
 
 ictv <- read_csv("Source/ICTV Master Species List 2019.v1.csv")
 
-virion %<>% mutate(ICTVRatified = (Virus %in% str_to_lower(ictv$Species)))
+virion %<>% mutate(ICTVRatified = (Virus %in% str_to_lower(ictv$Species))) %>%
+  relocate(ICTVRatified, .after = VirusNCBIResolved)
+
+#### SOME TEMPORARY SCAFFOLDING: May 30, 2021
+#### From the SRA patch
+
+virion %<>% mutate(HostFlagID = replace_na(HostFlagID, "FALSE"))
+virion %<>% select(-c(HostSynonyms, VirusFlagContaminant))
+
+####
 
 virion %<>% distinct()
+virion %<>% mutate(across(everything(), ~replace_na(.x, '')))
 
 vroom_write(virion, "Virion/Virion.csv.gz")

Code/05_Dissolve_VIRION.R

@@ -14,12 +14,18 @@ fixer <- function(x) {toString(unique(unlist(x)))}
 virion %<>% filter(!is.na(HostTaxID),
                    !is.na(VirusTaxID))
 
+# virion %>% 
+#   select(HostTaxID, Host, HostGenus, HostFamily, HostOrder, HostClass, HostSynonyms, HostNCBIResolved) %>% 
+#   group_by_at(vars(-c("HostSynonyms"))) %>% 
+#   summarise_at(vars(c("HostSynonyms")), ~list(.x)) %>%
+#   arrange(Host) %>%
+#   mutate(HostSynonyms = sapply(HostSynonyms, fixer)) -> host.tax
+
+# Temporary no-synonyms code:
 virion %>% 
-  select(HostTaxID, Host, HostGenus, HostFamily, HostOrder, HostClass, HostSynonyms, HostNCBIResolved) %>% 
-  group_by_at(vars(-c("HostSynonyms"))) %>% 
-  summarise_at(vars(c("HostSynonyms")), ~list(.x)) %>%
-  arrange(Host) %>%
-  mutate(HostSynonyms = sapply(HostSynonyms, fixer)) -> host.tax
+  select(HostTaxID, Host, HostGenus, HostFamily, HostOrder, HostClass, HostNCBIResolved) %>% 
+  distinct() %>% 
+  arrange(Host) -> host.tax
 
 virion %>% 
   select(VirusTaxID, Virus, VirusGenus, VirusFamily, VirusOrder, VirusClass, VirusNCBIResolved, ICTVRatified) %>% 
@@ -32,7 +38,7 @@ write_csv(host.tax, "Virion/TaxonomyHost.csv")
 write_csv(virus.tax, "Virion/TaxonomyVirus.csv")
 
 virion %<>% 
-  select(-c(Host, HostNCBIResolved, HostGenus, HostFamily, HostOrder, HostClass, HostSynonyms, 
+  select(-c(Host, HostNCBIResolved, HostGenus, HostFamily, HostOrder, HostClass,
             Virus, VirusNCBIResolved, VirusGenus, VirusFamily, VirusOrder, VirusClass, ICTVRatified)) #
 
 # Organize the sampling information into an ID-linked column
@@ -51,7 +57,7 @@ virion %>%
 virion %>% 
   select(AssocID,
          DetectionMethod, DetectionOriginal, 
-         HostFlagID, VirusFlagContaminant,
+         HostFlagID,
          NCBIAccession) -> detection
 
 virion %>% 
@@ -70,7 +76,7 @@ virion %<>%
             PMID,
             DetectionMethod, DetectionOriginal, 
             HostOriginal, VirusOriginal,
-            HostFlagID, VirusFlagContaminant,
+            HostFlagID,
             NCBIAccession,
             PublicationYear, 
             ReleaseYear, ReleaseMonth, ReleaseDay,

Code/Code_Dev/X_Geographic coverage.R

@@ -0,0 +1,99 @@
+
+setwd("~/Github/virion")
+library(tidyverse)
+library(vroom)
+library(sf)
+library(fasterize)
+
+vir <- vroom("Virion/virion.csv.gz")
+iucn <- st_read(dsn = 'C:/Users/cjcar/Dropbox/CurrentIUCN',
+                layer = 'MAMMALS')
+
+vir %>% filter(HostClass == "mammalia") %>%
+  select(Host, Virus) %>%
+  distinct() %>%
+  group_by(Host) %>%
+  summarize(NVirus = n_distinct(Virus)) -> nvir
+
+iucn %>% mutate(binomial = tolower(binomial)) %>%
+  left_join(nvir, by = c('binomial' = 'Host')) %>%
+  filter(NVirus > 0) -> iucn
+
+r <- raster::getData("worldclim",var="alt",res=5) # Make a blank raster
+map.num.m <- fasterize(iucn, r, field = NULL, fun = 'count')
+map.sum.m <- fasterize(iucn, r, field = "NVirus", fun = 'sum')
+
+############
+
+iucn <- st_read(dsn = 'C:/Users/cjcar/Dropbox/CurrentIUCN',
+                layer = 'REPTILES')
+
+vir %>% filter(HostClass == "lepidosauria" | HostOrder == "testudines") %>%
+  select(Host, Virus) %>%
+  distinct() %>%
+  group_by(Host) %>%
+  summarize(NVirus = n_distinct(Virus)) -> nvir
+
+iucn %>% mutate(binomial = tolower(binomial)) %>%
+  left_join(nvir, by = c('binomial' = 'Host')) %>%
+  filter(NVirus > 0) -> iucn
+
+r <- raster::getData("worldclim",var="alt",res=5) # Make a blank raster
+map.num.r <- fasterize(iucn, r, field = NULL, fun = 'count')
+map.sum.r <- fasterize(iucn, r, field = "NVirus", fun = 'sum')
+
+############
+
+iucn <- st_read(dsn = 'C:/Users/cjcar/Dropbox/CurrentIUCN',
+                layer = 'AMPHIBIANS')
+
+vir %>% filter(HostClass == "amphibia") %>%
+  select(Host, Virus) %>%
+  distinct() %>%
+  group_by(Host) %>%
+  summarize(NVirus = n_distinct(Virus)) -> nvir
+
+iucn %>% mutate(binomial = tolower(binomial)) %>%
+  left_join(nvir, by = c('binomial' = 'Host')) %>%
+  filter(NVirus > 0) -> iucn
+
+r <- raster::getData("worldclim",var="alt",res=5) # Make a blank raster
+map.num.a <- fasterize(iucn, r, field = NULL, fun = 'count')
+map.sum.a <- fasterize(iucn, r, field = "NVirus", fun = 'sum')
+
+############
+
+library(rgdal)
+bl <- st_read(dsn="C:/Users/cjcar/Dropbox/Dan\'s birdlife hole/BOTW.gdb",layer="All_Species")
+
+vir %>% filter(HostClass == "aves") %>%
+  select(Host, Virus) %>%
+  distinct() %>%
+  group_by(Host) %>%
+  summarize(NVirus = n_distinct(Virus)) -> nvir
+
+bl %>% mutate(SCINAME = tolower(SCINAME)) %>%
+  left_join(nvir, by = c('SCINAME' = 'Host')) %>%
+  filter(NVirus > 0) -> bl2
+
+map.num.b <- fasterize(bl2, r, field = NULL, fun = 'count')
+map.sum.b <- fasterize(bl2, r, field = "NVirus", fun = 'sum')
+
+library(rasterVis)
+
+s1 <- stack(map.num.m, map.num.b, map.num.r, map.num.a)
+names(s1) <- c("Mammals", "Birds", "Reptiles", "Amphibians")
+levelplot(log(s1+1), names.attr = names(s1))
+
+s2 <- stack(map.sum.m, map.sum.b, map.sum.r, map.sum.a)
+names(s2) <- c("Mammals", "Birds", "Reptiles", "Amphibians")
+levelplot(log(s2+1), names.attr = names(s2))
+
+s3 <- stack(map.num.m, map.sum.m,
+            map.num.b, map.sum.b,
+            map.num.r, map.sum.r,
+            map.num.a, map.sum.a)
+names(s3) <- c("Mammal (species)", "Mammal (interactions)",
+               "Bird (species)", "Bird (interactions)",
+               "Reptile (species)", "Reptile (interactions)",
+               "Amphibian (species)", "Amphibian (interactions)")