library(tidyr)
library(dplyr)
library(ggplot2)
library(reshape2)

#Please modify the path to match the directory on your own computer before running the script.
setwd("/Users/Kimashi/Documents/ws/tunome/BlastHomolog")

#All CSV files should be placed inside the DName folder directly under the specified path above.
DName="BER"

## Extract gene names from filenames
files=list.files(DName)
files2=files[grepl("^BlastCompTable_",files)]
files3=gsub("BlastCompTable_","",files2)
genes=gsub(".csv","",files3)

# Test loading one table
df <- read.table(paste0(DName,"/BlastCompTable_",genes[1],".csv"), header = TRUE, stringsAsFactors = FALSE, sep=",")
# Get species list
splist=unique(df$SPECIES[df$SPECIES!=""])

# Create result table
result <- setNames(data.frame(matrix(ncol = length(splist)+1, nrow = length(genes))), c("GeneName",splist))
result$GeneName=genes

for(i in seq_along(genes)){
  df <- read.table(paste0(DName,"/BlastCompTable_",genes[i],".csv"), header = TRUE, stringsAsFactors = FALSE, sep=",")
  # Fill empty species names with previous value
  df_filled <- df %>%mutate(SPECIES = ifelse(SPECIES == "", NA, SPECIES)) %>%fill(SPECIES)
  # Extract LOCUS (remove variant suffix)
  df_filled$LOCUS <- sub("\\.[^.]+$", "", df_filled$SSEQID)
  # Remove duplicated variants
  df_unique <- df_filled[!duplicated(df_filled[,c("SPECIES","LOCUS")]), ]
  # Remove NA rows
  df_omit<-na.omit(df_unique)
  
  for(j in seq_along(splist)){
    result[i,splist[j]]=sum(df_omit$SPECIES==splist[j])  
  }
}

# Convert to matrix
mat <- as.matrix(result[, -1])
rownames(mat) <- result$GeneName

# Format species names
colnames(mat) <- sub("^(.)(.)", "\\1. \\2", colnames(mat))
colnames(mat)[colnames(mat) == "P. egea"] <- "Pegea sp."

# Normalize each row by the maximum value
row_max <- apply(mat, 1, max)
row_max[row_max == 0] <- 1
mat_scaled <- mat / row_max
##########################
###########GGplot2#######
##########################
mat=mat_scaled

# Convert matrix to long format
mat_long <- melt(mat)
colnames(mat_long) <- c("Gene", "Species", "Value")

# Flag zero values
mat_long$is_zero <- mat_long$Value == 0

# Draw heatmap
g<-ggplot(mat_long, aes(x = Species, y = Gene, fill = Value)) +
  geom_tile(color = "grey50", size = 0.3) +
  geom_tile(data = mat_long[mat_long$is_zero, ],
            aes(x = Species, y = Gene), 
            color = "black", size = 0.4, fill = NA) +  # 0 のセルに太枠だけ描く
  scale_fill_gradient(low = "white", high = "red") +
  theme_minimal(base_size = 10) +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, face="italic"))

# Save figure
g
ggsave(paste0(DName,"/",DName,".png"), plot = g)