coolneng
/
locigenesis
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove redundant sequencing runs argument

This commit is contained in:
coolneng 2021-03-29 20:30:16 +02:00
parent 13f453718d
commit 04415310c6
Signed by: coolneng
GPG Key ID: 9893DA236405AF57
2 changed files with 12 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
generation.sh
View File

@ -20,5 +20,5 @@ filename="sequence"
prefix="curesim_"
Rscript src/repertoire.r "$sequences" "$sequencing_runs"
java -jar tools/CuReSim.jar -m "$read_mean_size" -sd "$read_variance_size" -f "$data_directory$filename$fasta" -o "$data_directory$prefix$filename$fastq"
java -jar tools/CuReSim.jar -n "$sequencing_runs" -m "$read_mean_size" -sd "$read_variance_size" -f "$data_directory$filename$fasta" -o "$data_directory$prefix$filename$fastq"
rm "$data_directory/log.txt"

40
src/repertoire.r
View File

@ -10,44 +10,30 @@ generate_repertoire <- function(number_of_sequences) {
))
}
amplify_rows <- function(data, column, factor) {
if (column == "sequence") {
dna_string <- Biostrings::DNAStringSet(data)
reverse_complement <- Biostrings::reverseComplement(dna_string)
return(rep(reverse_complement, factor))
}
return(rep(data, factor))
}
save_data <- function(data) {
Biostrings::writeXStringSet(data$sequence, "data/sequence.fasta")
vdj_sequences <- data[-1]
write.csv(vdj_sequences, "data/vdj_alignment.csv", row.names = FALSE)
}
process_data <- function(repertoire, sequencing_runs) {
process_data <- function(repertoire) {
columns <- c(
"sequence", "v_sequence_alignment",
"d_sequence_alignment", "j_sequence_alignment"
)
data <- repertoire[, columns]
amplified_data <- mapply(data, names(data),
sequencing_runs,
FUN = amplify_rows
)
save_data(amplified_data)
}
parse_cli_arguments <- function(args) {
if (length(args) != 2) {
stop("usage: repertoire.r <number of sequences> <sequencing_runs>")
}
return(c(args[1], args[2]))
data$sequence <- Biostrings::reverseComplement(data$sequence)
save_data(data)
}
parse_cli_arguments <- function() {
args <- commandArgs(trailingOnly = TRUE)
arguments <- parse_cli_arguments(args)
number_of_sequences <- as.integer(arguments[1])
sequencing_runs <- as.integer(arguments[2])
repertoire <- generate_repertoire(number_of_sequences)
process_data(repertoire, sequencing_runs)
if (length(args) != 1) {
stop("usage: repertoire.r <number of sequences>")
}
return(args[1])
}
arguments <- parse_cli_arguments(commandArgs(trailing))
repertoire <- generate_repertoire(number_of_sequences = arguments[1])
process_data(repertoire)