Merge pull request #17 from infinity-a11y/version_1.4.1

Version 1.4.1

PhyloTrace.yml

@@ -12,10 +12,10 @@ dependencies:
   - r-bh=1.81.0-1
   - r-biocmanager=1.30.22
   - bioconductor-biostrings=2.70.1
-  - bioconductor-ggtree=3.8.0
-  - bioconductor-ggtreeextra=1.10.0
+  - bioconductor-ggtree=3.10.0
+  - bioconductor-ggtreeextra=1.12.0
   - bioconductor-biocversion=3.17.1
-  - bioconductor-treeio=1.24.1
+  - bioconductor-treeio=1.26.0
   - r-dbi=1.1.3
   - r-mass=7.3_60
   - r-matrix=1.6_1.1
@@ -98,6 +98,7 @@ dependencies:
   - r-lattice=0.22_5
   - r-lazyeval=0.2.2
   - r-lifecycle=1.0.4
+  - r-logr=1.3.8
   - r-lubridate=1.9.3
   - r-magrittr=2.0.3
   - r-memoise=2.0.1
@@ -137,7 +138,8 @@ dependencies:
   - r-sass=0.4.9
   - r-scales=1.2.1
   - r-selectr=0.4_2
-  - r-shiny=1.7.5
+  - r-shiny=1.8.1.1
+  - r-shinybs=0.61.1
   - r-shinyFiles=0.9.3
   - r-shinyWidgets=0.8.0
   - r-shinydashboard=0.7.2

Report/Report.Rmd

@@ -44,20 +44,23 @@ knitr::include_graphics(paste0(getwd(),"/PhyloTrace_bw.png"))
 
 \
 
-```{r echo=FALSE, fig.align = "center"}
+```{r echo=FALSE, fig.align = "center", out.width='100%'}
 if(!my_loaded_variables[["plot"]] == "MST"){
   if(my_loaded_variables[["report_df.Include"]][13]) {
   knitr::include_graphics(paste0(my_loaded_variables[["plot"]], ".jpeg"))
   }
+} else {
+  if(my_loaded_variables[["report_df.Include"]][13]) {
+  knitr::include_graphics(paste0(my_loaded_variables[["plot"]], ".jpg"))
+  }
 }
-
 ```
 
 \
 
 ```{r echo=FALSE}
 if(my_loaded_variables[["report_df.Include"]][1]) {
-  kable(select(my_loaded_variables[["entry_table"]], -2), 
+  kable(select(my_loaded_variables[["entry_table"]], my_loaded_variables[["table_columns"]]), 
       row.names = FALSE,
       linesep = "") %>%
   kable_styling(bootstrap_options = "striped", full_width = TRUE, font_size = 10) %>%
@@ -90,7 +93,7 @@ if(!is.null(my_loaded_variables[["na_handling"]])){
                           if(my_loaded_variables[["report_df.Include"]][10]){"Distance algorithm"},
                           if(my_loaded_variables[["report_df.Include"]][11]){"Missing value handling"},
                           if(my_loaded_variables[["report_df.Include"]][12]){"PhyloTrace version"},
-                          if(my_loaded_variables[["report_df.Include"]][12]){"KMA version"}),
+                          if(my_loaded_variables[["report_df.Include"]][12]){"pblat version"}),
                  Content = c(if(my_loaded_variables[["report_df.Include"]][9]){my_loaded_variables[["tree"]]},
                              if(my_loaded_variables[["report_df.Include"]][10]){my_loaded_variables[["distance"]]},
                              if(my_loaded_variables[["report_df.Include"]][11]){my_loaded_variables[["na_handling"]]},
@@ -101,7 +104,7 @@ if(!is.null(my_loaded_variables[["na_handling"]])){
   df <- data.frame(Item = c(if(my_loaded_variables[["report_df.Include"]][9]){"Tree construction"},
                           if(my_loaded_variables[["report_df.Include"]][10]){"Distance algorithm"},
                           if(my_loaded_variables[["report_df.Include"]][12]){"PhyloTrace version"},
-                          if(my_loaded_variables[["report_df.Include"]][12]){"KMA version"}),
+                          if(my_loaded_variables[["report_df.Include"]][12]){"pblat version"}),
                  Content = c(if(my_loaded_variables[["report_df.Include"]][9]){my_loaded_variables[["tree"]]},
                              if(my_loaded_variables[["report_df.Include"]][10]){my_loaded_variables[["distance"]]},
                              if(my_loaded_variables[["report_df.Include"]][12]){my_loaded_variables[["version"]][1]},

execute/check_duplicate_multi.R

@@ -1,3 +1,4 @@
+library(logr)
 # Get the command line arguments
 args <- commandArgs(trailingOnly = TRUE)
 
@@ -12,8 +13,11 @@ log.message <- function(log_file, message) {
   cat(format(Sys.time(), "%Y-%m-%d %H:%M:%S"), "-", message, "\n", file = log_file, append = TRUE)
 }
 
-log.message(log_file = paste0(base_path, "/logs/output.log"),
-            message = "Initiated multi typing fasta name duplicates check")
+logfile <- file.path(paste0(base_path, "/logs/check_duplicate_multi.log"))
+
+log <- log_open(logfile, logdir = FALSE)
+
+log_print("Initiated multi typing fasta name duplicates check")
 
 # load selected assemblies
 assemblies <- lapply(list.files(paste0(getwd(), "/selected_genomes"), full.names = T), readLines)
@@ -25,12 +29,12 @@ for(i in 1:length(assemblies)){
   # Test if there are duplicates
   if(length(names) != length(unique(names))){
 
-    log.message(log_file = paste0(base_path, "/logs/output.log"),
-                message = paste0("Duplicate(s) present in ", basename(file_names[i])))
+    log_print(paste0("Duplicate(s) present in ", basename(file_names[i])))
 
     # add a number to the duplicates
     for(j in 1:length(names)){
       if(sum(names == names[j]) > 1){
+        indices <- which(names == names[j])
         names[j] <- paste0(names[j], "_", which(names == names[j]))
       }
     }
@@ -44,3 +48,5 @@ for(i in 1:length(assemblies)){
     writeLines(assemblies[[i]], file_names[i])
   } 
 }
+
+log_close()

execute/check_duplicate.R → execute/check_duplicate_single.R

@@ -1,12 +1,16 @@
+library(logr)
 typing_meta <- readRDS(paste0(getwd(), "/single_typing_df.rds"))
 
 # Function to log messages 
 log.message <- function(log_file, message) {
   cat(format(Sys.time(), "%Y-%m-%d %H:%M:%S"), "-", message, "\n", file = log_file, append = TRUE)
 }
 
-log.message(log_file = paste0(typing_meta$wd, "/logs/output.log"),
-            message = "Initiated single typing fasta name duplicates check")
+logfile <- file.path(paste0(typing_meta$wd, "/logs/check_duplicate_single.log"))
+
+log <- log_open(logfile, logdir = FALSE)
+
+log_print("Initiated single typing fasta name duplicates check")
 
 assembly <- typing_meta$genome
 
@@ -15,20 +19,20 @@ lines <- readLines(assembly)
 names <- stringr::str_extract(lines[seq(1, length(lines), by = 3)], "^[^\\s]+")
 
 # Test if there are duplicates
-if(length(names) != length(unique(names))){
+if(length(names) != length(unique(names))) {
 
-  log.message(log_file = paste0(typing_meta$wd, "/logs/output.log"),
-              message = paste0("Duplicate(s) present in ", basename(assembly)))
+  log_print(paste0("Duplicate(s) present in ", basename(assembly)))
 
   # add a number to the duplicates
-  for(i in 1:length(names)){
-    if(sum(names == names[i]) > 1){
-      names[i] <- paste0(names[i], "_", which(names == names[i]))
+  for(i in 1:length(names)) {
+    if(sum(names == names[i]) > 1) {
+      indices <- which(names == names[i])
+      names[i] <- paste0(names[i], "_", indices[which(indices == i)])
     }
   }
 
   # substitute the respective lines in the file with the new names 
-  for(i in 1:length(names)){
+  for(i in 1:length(names)) {
     lines[3*i - 2] <- paste0(names[i])
   }
 
@@ -37,3 +41,5 @@ if(length(names) != length(unique(names))){
 } else {
   writeLines(lines, paste0(getwd(), "/blat_single/assembly.fasta"))
 }
+
+log_close()

execute/kill_multi.sh

@@ -2,8 +2,8 @@
 unset R_HOME
 
 log_file='execute/script_log.txt'
-blat_multi='execute/blat_multi.sh'
-automatic_typing='automatic_typing.R'
+blat_multi='execute/multi_typing.sh'
+automatic_typing='multi_eval.R'
 
 # Function to log messages to the file
 log_message() {

execute/automatic_typing.R → execute/multi_eval.R

@@ -1,3 +1,5 @@
+library(logr)
+
 meta_info <- readRDS("meta_info.rds")
 db_path <- readRDS("multi_typing_df.rds")[, "db_path"]
 assembly_folder <- dir(paste0(getwd(), "/selected_genomes"), full.names = TRUE)
@@ -26,8 +28,11 @@ log.message <- function(log_file, message) {
   cat(format(Sys.time(), "%Y-%m-%d %H:%M:%S"), "-", message, "\n", file = log_file, append = TRUE)
 }
 
-log.message(log_file = paste0(meta_info$db_directory, "/logs/output.log"),
-            message = "Attaching initiated (automatic_typing.R)")
+logfile <- file.path(paste0(getwd(), "/logs/multi_eval.log"))
+
+log <- log_open(logfile, logdir = FALSE)
+
+log_print("Attaching initiated")
 
 # Define start and stop codons
 start_codons <- c("ATG", "GTG", "TTG")
@@ -309,18 +314,18 @@ if(sum(unname(base::sapply(psl_files, file.size)) <= 427) / length(psl_files) <=
   saveRDS(Database, paste0(db_path, "/", gsub(" ", "_", meta_info$cgmlst_typing), "/Typing.rds"))
 
   # Logging successes
-  log.message(log_file = paste0(getwd(), "/execute/script_log.txt"), 
-              message = paste0("Successful typing of ", sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
-  log.message(log_file = paste0(meta_info$db_directory, "/logs/output.log"), 
+  log.message(log_file = paste0(getwd(), "/logs/script_log.txt"), 
               message = paste0("Successful typing of ", sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
+  log_print(paste0("Successful typing of ", sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
 
 } else {
 
   # Logging failures
-  log.message(log_file = paste0(getwd(), "/execute/script_log.txt"), 
-              message = paste0("Assembly typing failed for ", 
-                               sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
-  log.message(log_file = paste0(meta_info$db_directory, "/logs/output.log"), 
+  log.message(log_file = paste0(getwd(), "/logs/script_log.txt"), 
               message = paste0("Assembly typing failed for ", 
                                sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
+  log_print(paste0("Assembly typing failed for ", 
+                         sub("\\.(fasta|fna|fa)$", "", basename(assembly))))
 }
+
+log_close()

execute/blat_multi.sh → execute/multi_typing.sh

@@ -26,10 +26,7 @@ fi
 mkdir "$results"
 
 selected_genomes="$base_path/execute/selected_genomes"
-log_file="$base_path/execute/script_log.txt"
-output_file="$base_path/logs/output.log"
-
-echo "$(date +"%Y-%m-%d %H:%M:%S") - Initiated blat_multi.sh" >> "$output_file"
+log_file="$base_path/logs/script_log.txt"
 
 # Create a log file or truncate if it exists
 echo "Start Multi Typing with $scheme scheme." > "$log_file"
@@ -47,10 +44,8 @@ for file in "${file_names[@]}"; do
     if [ -f "$genome_folder/$file" ]; then
         cp "$genome_folder/$file" "$selected_genomes/"
         echo "$(date +"%Y-%m-%d %H:%M:%S") - Initiated $file" >> "$log_file"
-        echo "$(date +"%Y-%m-%d %H:%M:%S") - Initiated $file" >> "$output_file"
     else
         echo "$(date +"%Y-%m-%d %H:%M:%S") - $file not found in $genome_folder" >> "$log_file"
-        echo "$(date +"%Y-%m-%d %H:%M:%S") - $file not found in $genome_folder" >> "$output_file"
     fi
 done
 
@@ -70,18 +65,15 @@ for genome in "$selected_genomes"/*; do
     genome_filename=$(basename "$genome")
     genome_filename_noext="${genome_filename%.*}"
     echo "$(date +"%Y-%m-%d %H:%M:%S") - Processing $genome_filename" >> "$log_file"
-    echo "$(date +"%Y-%m-%d %H:%M:%S") - Processing $genome_filename" >> "$output_file"
     fi
     mkdir "$results/$genome_filename_noext"
 
     result_folder="$results/$genome_filename_noext"
 
     # Run parallelized BLAT
-    find "$alleles" -type f \( -name "*.fasta" -o -name "*.fa" -o -name "*.fna" \) | parallel pblat $genome {} "$result_folder/{/.}.psl"
+    find "$alleles" -type f \( -name "*.fasta" -o -name "*.fa" -o -name "*.fna" \) | parallel pblat $genome {} "$result_folder/{/.}.psl"  > /dev/null 2>&1
 
     echo "$(date +"%Y-%m-%d %H:%M:%S") - Attaching $genome_filename" >> "$log_file"
-    echo "$(date +"%Y-%m-%d %H:%M:%S") - Initiating addition of $genome_filename (attaching)" >> "$output_file"
-    Rscript "$base_path/execute/automatic_typing.R" "$genome_filename"
+    Rscript "$base_path/execute/multi_eval.R" "$genome_filename"
 done
 echo "$(date +"%Y-%m-%d %H:%M:%S") - Multi Typing finalized." >> "$log_file"
-echo "$(date +"%Y-%m-%d %H:%M:%S") - Multi Typing finalized." >> "$output_file"

execute/single_typing.R → execute/single_eval.R

@@ -1,3 +1,5 @@
+library(logr)
+
 # Hand over variables
 meta_info <- readRDS("meta_info_single.rds")
 db_path <- readRDS("single_typing_df.rds")[, "db_path"]
@@ -25,8 +27,11 @@ log.message <- function(log_file, message) {
   cat(format(Sys.time(), "%Y-%m-%d %H:%M:%S"), "- ", message, "\n", file = log_file, append = TRUE)
 }
 
-log.message(log_file = paste0(meta_info$db_directory, "/logs/output.log"),
-            message = "Attaching initiated (single_typing.R)")
+logfile <- file.path(paste0(getwd(), "/logs/single_eval.log"))
+
+log <- log_open(logfile, logdir = FALSE)
+
+log_print("Attaching initiated")
 
 # Define start and stop codons
 start_codons <- c("ATG", "GTG", "TTG")
@@ -142,34 +147,27 @@ if(sum(unname(base::sapply(psl_files, file.size)) <= 427) / length(psl_files) <=
 
     Database <- list(Typing = data.frame())
 
-    Typing <-
-      data.frame(matrix(
-        NA,
-        nrow = 0,
-        ncol = 12 + length(psl_files)
-      ))
+    Typing <- data.frame(matrix(NA, nrow = 0, ncol = 12 + length(psl_files)))
 
-    metadata <-
-      c(
-        1,
-        TRUE,
-        meta_info$assembly_id,
-        meta_info$assembly_name,
-        meta_info$cgmlst_typing,
-        as.character(meta_info$append_isodate),
-        meta_info$append_host,
-        meta_info$append_country,
-        meta_info$append_city,
-        as.character(meta_info$append_analysisdate),
-        length(allele_vector) - sum(sapply(allele_vector, is.na)),
-        sum(sapply(allele_vector, is.na))
-      )
+    metadata <- c(
+      1,
+      TRUE,
+      meta_info$assembly_id,
+      meta_info$assembly_name,
+      meta_info$cgmlst_typing,
+      as.character(meta_info$append_isodate),
+      meta_info$append_host,
+      meta_info$append_country,
+      meta_info$append_city,
+      as.character(meta_info$append_analysisdate),
+      length(allele_vector) - sum(sapply(allele_vector, is.na)),
+      sum(sapply(allele_vector, is.na))
+    )
 
     new_row <- c(metadata, allele_vector)
 
     Typing <- rbind(Typing, new_row)
 
-
     colnames(Typing) <-
       append(
         c(
@@ -298,18 +296,18 @@ if(sum(unname(base::sapply(psl_files, file.size)) <= 427) / length(psl_files) <=
   saveRDS(Database, paste0(db_path, "/", gsub(" ", "_", meta_info$cgmlst_typing), "/Typing.rds"))
 
   # Logging successes
-  log.message(log_file = paste0(getwd(), "/execute/single_typing_log.txt"), 
-              message = paste0("Successful typing of ", meta_info$assembly_name))
-  log.message(log_file = paste0(getwd(), "/logs/output.log"), 
+  log.message(log_file = paste0(getwd(), "/logs/single_typing_log.txt"), 
               message = paste0("Successful typing of ", meta_info$assembly_name))
+  log_print(paste0("Successful typing of ", meta_info$assembly_name))
 
 } else {
 
   failures <- sum(unname(base::sapply(psl_files, file.size)) <= 100) / length(psl_files) * 100
 
   # Logging failures
-  log.message(log_file = paste0(getwd(), "/execute/single_typing_log.txt"), 
-              message = paste0("Assembly typing of ", meta_info$assembly_name, " failed. ", failures, "% of loci not typed."))
-  log.message(log_file = paste0(getwd(), "/logs/output.log"), 
+  log.message(log_file = paste0(getwd(), "/logs/single_typing_log.txt"), 
               message = paste0("Assembly typing of ", meta_info$assembly_name, " failed. ", failures, "% of loci not typed."))
+  log_print(paste0("Assembly typing of ", meta_info$assembly_name, " failed. ", failures, "% of loci not typed."))
 }
+
+log_close()

execute/blat_run.sh → execute/single_typing.sh

@@ -35,20 +35,17 @@ fi
 mkdir "$results"
 
 # Check assembly file and save in the execute folder
-Rscript "$base_path/execute/check_duplicate.R"
+Rscript "$base_path/execute/check_duplicate_single.R"
 wait
 genome="$base_path/execute/blat_single/assembly.fasta"
 
 # Run parallelized BLAT
-parallel --citation
-find "$alleles" -type f \( -name "*.fasta" -o -name "*.fa" -o -name "*.fna" \) | parallel pblat $genome {} "$results/{/.}.psl"
+find "$alleles" -type f \( -name "*.fasta" -o -name "*.fa" -o -name "*.fna" \) | parallel pblat $genome {} "$results/{/.}.psl" > /dev/null 2>&1
 
 # Start appending results
 echo 888888 >> "$base_path/logs/progress.txt"
-echo "Initiating addition of $genome_name (attaching)" >> "$base_path/logs/output.log"
-Rscript "$base_path/execute/single_typing.R"
+Rscript "$base_path/execute/single_eval.R"
 wait
 
 # Single typing finalized
 echo 999999 >> "$base_path/logs/progress.txt"
-echo "Finished typing of $genome_name" >> "$base_path/logs/output.log"