enrichment_sxbd = read.table("~/KEGGTOP20.txt", header=T, row.names=NULL,
                        sep="\t", quote="")
head(enrichment_sxbd)

library(plyr)
library(stringr)
library(ggplot2)
library(grid)
mixedToFloat <- function(x){
  x <- sapply(x, as.character)
  is.integer <- grepl("^-?\\d+$", x)
  is.fraction <- grepl("^-?\\d+\\/\\d+$", x)
  is.float <- grepl("^-?\\d+\\.\\d+$", x)
  is.mixed <- grepl("^-?\\d+ \\d+\\/\\d+$", x)
  stopifnot(all(is.integer | is.fraction | is.float | is.mixed))
  numbers <- strsplit(x, "[ /]")
  ifelse(is.integer, as.numeric(sapply(numbers, `[`, 1)),
         ifelse(is.float, as.numeric(sapply(numbers, `[`, 1)),
                ifelse(is.fraction, as.numeric(sapply(numbers, `[`, 1)) /
                         as.numeric(sapply(numbers, `[`, 2)),
                       as.numeric(sapply(numbers, `[`, 1)) +
                         as.numeric(sapply(numbers, `[`, 2)) /
                         as.numeric(sapply(numbers, `[`, 3)))))
}
mixedToFloat(c('1 1/2', '2 3/4', '2/3', '11 1/4', '1'))

enrichment_sxbd$GeneRatio = mixedToFloat(enrichment_sxbd$GeneRatio)
enrichment_sxbd$Count = mixedToFloat(enrichment_sxbd$Count)


log_name = paste0("negLog10_", "pvalue")
col_name_enrichment_sxbd <- colnames(enrichment_sxbd)
col_name_enrichment_sxbd <- c(col_name_enrichment_sxbd, log_name)
enrichment_sxbd$log_name <- log10(enrichment_sxbd$pvalue) * (-1)
colnames(enrichment_sxbd) <- col_name_enrichment_sxbd


enrichment_sxbd_freq <- as.data.frame(table(enrichment_sxbd$Description))
colnames(enrichment_sxbd_freq) <- c("Description", "IDctct")
head(enrichment_sxbd_freq)


enrichment_sxbd2 <- merge(enrichment_sxbd, enrichment_sxbd_freq, by="Description")


enrichment_sxbd3 <- enrichment_sxbd2[order(enrichment_sxbd2$IDctct,
                                           enrichment_sxbd2$GeneRatio, enrichment_sxbd2$negLog10_pvalue), ]
term_order <- unique(enrichment_sxbd3$Description)


enrichment_sxbd$Description <- factor(enrichment_sxbd$Description,
                                      levels=term_order, ordered=T)
color_v <- c("green", "red")


p <- ggplot(enrichment_sxbd, aes(x=GeneRatio,y=Description)) +
  labs(x="GeneRatio (%)", y="KEGG Pathway") + labs(title="")
p <- p + geom_point(aes(size=Count, color=negLog10_pvalue )) +
  scale_colour_gradient(low=color_v[1], high=color_v[2], name="negLog10_pvalue")

p <- p + scale_y_discrete(labels=function(x) str_wrap(x, width=60))
p <- p + theme_bw() +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())
top='top'
bottom='bottom'
left='left'
right='right'
none='none'
legend_pos_par <- right
uwid = 0
vhig = 12

if (uwid == 0 || vhig == 0) {
  x_len = length(unique(enrichment_sxbd$Description))
  if(x_len<10){
    vhig = 10
  } else if(x_len<20) {
    vhig = 10 + (x_len-10)/3
  } else if(x_len<100) {
    vhig = 13 + (x_len-20)/5
  } else {
    vhig = 40
  }
  uwid = vhig
  if(legend_pos_par %in% c("left", "right")){
    uwid = 1.5 * uwid
  }
}
p <- p + theme(legend.position=legend_pos_par)
p <- p + theme( panel.grid = element_blank(), panel.border=element_blank(),
                legend.background = element_blank(),
                axis.line.x=element_line(size=0.4, colour="black", linetype='solid'),
                axis.line.y=element_line(size=0.4, colour="black", linetype='solid'),
                axis.ticks = element_line(size=0.4)
)
ggsave(p, filename="LUADKEGG20.pdf", dpi=300, width=uwid,
height=vhig, units=c("cm"))
p