36 lines
1.4 KiB
R
36 lines
1.4 KiB
R
|
#!/usr/bin/env Rscript
|
||
|
|
suppressPackageStartupMessages({
|
||
|
|
library(argparser, quietly = TRUE)
|
||
|
|
library(ggplot2)
|
||
|
|
library(ggseqlogo)
|
||
|
|
})
|
||
|
|
parser <- arg_parser("draw motif seqlogo for meme result.")
|
||
|
|
parser <- add_argument(parser, "input", help = "input file, protein motif sequence table", type = "character")
|
||
|
|
parser <- add_argument(parser, "outpre", help = "output figure prefix", type = "character")
|
||
|
|
argv <- parse_args(parser)
|
||
|
|
data <- read.table(file = argv$input, header = FALSE, stringsAsFactors = FALSE)
|
||
|
|
motif_names <- unique(data[, 5])
|
||
|
|
all_motif_list <- list()
|
||
|
|
for (motif in motif_names) {
|
||
|
|
seqs <- as.character(data[data[, 5] == motif, 4])
|
||
|
|
seq_lengths <- nchar(seqs)
|
||
|
|
if (length(unique(seq_lengths)) > 1) {
|
||
|
|
min_len <- min(seq_lengths)
|
||
|
|
message("Notice: Motif ", motif, " has different sequence lengths; trimming to ", min_len)
|
||
|
|
seqs <- substr(seqs, 1, min_len)
|
||
|
|
}
|
||
|
|
all_motif_list[[motif]] <- seqs
|
||
|
|
pdf(paste(argv$outpre, motif, "seqlogo.pdf", sep = "."), width = 15, height = 5)
|
||
|
|
print(ggseqlogo(seqs) + theme_logo())
|
||
|
|
dev.off()
|
||
|
|
message("Finished motif: ", motif)
|
||
|
|
}
|
||
|
|
if (length(all_motif_list) > 0) {
|
||
|
|
plot_height <- length(all_motif_list) * 1.8
|
||
|
|
pdf(paste(argv$outpre, "allmotif.seqlogo.pdf", sep = "."), width = 15, height = max(4, plot_height))
|
||
|
|
print(ggseqlogo(all_motif_list, ncol = 1) + theme_logo())
|
||
|
|
dev.off()
|
||
|
|
} else {
|
||
|
|
message("Warning: no motif sequences found; skipped combined plot")
|
||
|
|
}
|