From 8cfd107a2e1930a7447b503df053019677a2fa65 Mon Sep 17 00:00:00 2001
From: sdgamboa <samuel.gamboa.tuz@gmail.com>
Date: Tue, 9 Jan 2024 22:08:48 -0500
Subject: [PATCH] Update vignettes with new makeSignatures function and
 bugphyzz import.

---
 vignettes/articles/attributes.Rmd | 46 +++++++----------------
 vignettes/bugphyzz.Rmd            | 62 +++++++++++++++++--------------
 2 files changed, 49 insertions(+), 59 deletions(-)

diff --git a/vignettes/articles/attributes.Rmd b/vignettes/articles/attributes.Rmd
index 3d962e9d..4a724a3f 100644
--- a/vignettes/articles/attributes.Rmd
+++ b/vignettes/articles/attributes.Rmd
@@ -23,19 +23,20 @@ knitr::opts_chunk$set(
 library(DT)
 library(bugphyzz)
 library(dplyr)
+library(purrr)
 ```
 
 ```{r}
-bp <- importBugphyzz(version = 'devel', force_download = TRUE)
-bp <- bp |> 
-  filter(
-    !Evidence %in% c('asr', 'inh')
-  )
-x <- bp |> 
-  select(Attribute_source, Attribute_group, Attribute) |> 
-  filter(!grepl(';', Attribute_source)) |> 
-  count(Attribute_source, Attribute_group) |> 
-  mutate(new_col = paste0(Attribute_group, ' (', n, ')'))
+bp <- importBugphyzz(version = 'devel')
+x <- map(bp, ~ {
+  .x |> 
+    select(Attribute_group, Attribute, Attribute_source) |> 
+    count(Attribute_group, Attribute, Attribute_source, name = "N_annotations") |> 
+    distinct()
+}) |> 
+  bind_rows() |> 
+  filter(!is.na(Attribute_source)) |> # Attribute source with NAs means it was obtained through ASR or TAX
+  relocate(Attribute_source, Attribute_group, Attribute, N_annotations)
 ```
 
 ## Attributes
@@ -80,11 +81,6 @@ datt
 ## Sources
 
 ```{r}
-x <- x |> 
-  select(-Attribute_group, -n) |> 
-  group_by(Attribute_source) |> 
-  summarise(Attribute_group = paste(sort(new_col), collapse = '; ')) |> 
-  ungroup()
 fname2 <- system.file(
   'extdata/attribute_sources.tsv', 
   package = 'bugphyzz', mustWork = TRUE
@@ -93,23 +89,9 @@ src <- read.table(
   fname2, header = TRUE, sep = '\t', quote = ""
 )
 
-src <- left_join(x, src, by = 'Attribute_source') |> 
-  relocate(
-    Attribute_source, Confidence_in_curation, Evidence,
-    Attribute_group, full_source
-  ) |> 
-  mutate(
-    Confidence_in_curation = factor(
-      Confidence_in_curation, levels = c('high', 'medium', 'low')
-    )
-  ) |> 
-  arrange(Confidence_in_curation, Evidence, Attribute_source)
+xsrc <- left_join(x, src, by = 'Attribute_source') |> 
+  rename(Full_source = full_source)
   
-colnames(src) <- c(
-  'Source (short)', 'Confidence in curation', 'Evidence*', 'Attribute group**',
-  'Full source'
-)
-
 caption2 <- paste0(
   'Table 2. Sources of attribute annotations in bugphyzz. ',
   '* Evidence codes: exp = experimental evidence, igc = inferred from genomic context, ',
@@ -120,7 +102,7 @@ caption2 <- paste0(
 )
 
 src_dt <- datatable(
-  data = src, 
+  data = xsrc, 
   filter = "top",
   extensions = c("Buttons","KeyTable"), 
   # caption = caption2,
diff --git a/vignettes/bugphyzz.Rmd b/vignettes/bugphyzz.Rmd
index e8428f08..f5ecec60 100644
--- a/vignettes/bugphyzz.Rmd
+++ b/vignettes/bugphyzz.Rmd
@@ -22,7 +22,7 @@ also be accessed by direct download at **>>> insert Zenodo link here <<<**.
 
 # Import bugphyzz
 
-```{r, message=FALSE, eval=FALSE}
+```{r, eval=FALSE}
 if (!require("BiocManager", quietly = TRUE))
     install.packages("BiocManager")
 
@@ -37,62 +37,70 @@ library(purrr)
 ```
 
 ```{r import data}
-bp <- importBugphyzz(version = 'devel', force_download = TRUE)
-head(bp)
+bp <- importBugphyzz(version = 'devel')
+head(map(bp, head))
 ```
 
 ## Explore wich attribute groups are available
 
 ```{r}
-unique(bp$Attribute_group)
+names(bp)
 ```
 
-## Explore which attribute signatures are available
+# Create signatures
+
+## Create signatures of taxids at the genus level for aerophilicity
 
 ```{r}
-head(unique(bp$Attribute))
+aer_sigs_g <- makeSignatures(
+  dat = bp[["aerophilicity"]], tax_id_type = "NCBI_ID", tax_level = "genus"
+)
+map(aer_sigs_g, head)
 ```
 
-# Creating signatures
-
-## Create signatures of taxids at the genus level for aerophilicity
+## Create signatures of taxa names at the species level for growth temperature
 
 ```{r}
-aer <- bp[which(bp$Attribute_group == 'aerophilicity'),]
-sigs1 <- getBugphyzzSignatures(
-  df = aer, tax.id.type = 'NCBI_ID', tax.level = 'genus', min.size = 10
+gt_sigs_sp <- makeSignatures(
+  dat = bp[["growth temperature"]], tax_id_type = "Taxon_name",
+  tax_level = 'species'
 )
-map(sigs1, head)
+map(gt_sigs_sp, head)
 ```
 
-## Create signatures of taxa names at the species level for optimal ph 
+## Create signatures with custom threshold for numeric attributes
+
 ```{r}
-op <- bp[bp$Attribute_group == 'optimal ph', ]
-sigs2 <- getBugphyzzSignatures(
-  df = op, tax.id.type = 'Taxon_name', tax.level = 'species', min.size = 10
+gt_sigs_mix <- makeSignatures(
+  dat = bp[["growth temperature"]], tax_id_type = "Taxon_name",
+  tax_level = "mixed", min = 0, max = 25
 )
-map(sigs2, head)
+map(gt_sigs_mix, head)
 ```
 
-
-# Other examples
+## Create signatures for a binary attribute
 
 ```{r}
-mot <- bp[bp$Attribute_group == 'motility',]
-sigs3 <- getBugphyzzSignatures(
-  df = mot, tax.id.type = 'Taxon_name', tax.level = 'mixed'
+ap_sigs_mix <- makeSignatures(
+  dat = bp[["animal pathogen"]], tax_id_type = "NCBI_ID",
+  tax_level = "mixed", evidence = c("exp", "igc", "nas", "tas")
 )
-lapply(sigs3, head)
+map(ap_sigs_mix, head)
 ```
 
-# Merge examples above
+## Make signatures for all datasets with a single function call
 
 ```{r}
-sigs <- c(sigs1, sigs2, sigs3)
-lapply(sigs, head)
+sigs <- map(bp, makeSignatures) |> 
+  list_flatten()
+length(sigs)
 ```
 
 
+```{r}
+head(map(sigs, head))
+```
+
 ## Session information:
 
 ```{r}