Mercurial > repos > petr-novak > dante_ltr

annotate clean_ltr.R @ 2:f131886ea194 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 891bfe9acf7349c2b887aff6d7e52a7f4ebf3b3a"
author petr-novak
date Tue, 12 Apr 2022 12:55:32 +0000
parents 7b0bbe7477c4
children 6ae4a341d1f3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
1 #!/usr/bin/env Rscript
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
2 initial_options <- commandArgs(trailingOnly = FALSE)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
3 file_arg_name <- "--file="
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
4 script_name <- normalizePath(sub(file_arg_name, "", initial_options[grep
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
5 (file_arg_name,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
6 initial_options)]))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
7 script_dir <- dirname(script_name)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
8 library(optparse)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
9
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
10 parser <- OptionParser()
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
11 option_list <- list(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
12 make_option(c("-g", "--gff3"), action = "store", type = "character",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
13 help = "gff3 with LTR Transposable elements", default = NULL),
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
14 make_option(c("-s", "--reference_sequence"), action = "store", type = "character",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
15 help = "reference sequence as fasta",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
16 default = NULL),
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
17 make_option(c("-o", "--output"), action = "store", type = "character",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
18 help = "output file prefix", default = NULL),
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
19 make_option(c("-c", "--cpu"), type =
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
20 "integer", default = 5, help = "Number of cpu to use [default %default]",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
21 metavar = "number")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
22
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
23 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
24 description <- paste(strwrap(""))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
25
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
26 epilogue <- ""
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
27 parser <- OptionParser(option_list = option_list, epilogue = epilogue, description =
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
28 description, usage = "usage: %prog COMMAND [OPTIONS]")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
29 opt <- parse_args(parser, args = commandArgs(TRUE))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
30
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
31
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
32 # load packages
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
33 suppressPackageStartupMessages({ library(rtracklayer)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
34 library(Biostrings)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
35 library(BSgenome)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
36 library(parallel)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
37 })
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
38
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
39 # CONFIGURATION
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
40 # load configuration files and functions:
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
41 lineage_file <- paste0(script_dir, "/databases/lineage_domain_order.csv")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
42 ltr_utils_r <- paste0(script_dir, "/R/ltr_utils.R")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
43 if (file.exists(lineage_file)) {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
44 lineage_info <- read.table(lineage_file, sep = "\t",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
45 header = TRUE,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
46 as.is = TRUE)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
47 source(ltr_utils_r)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
48 }else {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
49 lineage_file <- paste0(script_dir, "/.
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
50 ./share/dante_ltr/databases/lineage_domain_order.csv")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
51 ltr_utils_r <- paste0(script_dir, "/.
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
52 ./share/dante_ltr/R/ltr_utils.R")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
53 if (file.exists(lineage_file)) {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
54 lineage_info <- read.table(lineage_file, sep = "\t",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
55 header = TRUE,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
56 as.is = TRUE)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
57 source(ltr_utils_r)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
58 }else {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
59 (stop("configuration files not found"))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
60 }
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
61 }
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
62
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
63
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
64 ncpus <- opt$cpu
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
65
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
66
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
67 # load data
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
68 cat("reading fasta...")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
69 s <- readDNAStringSet(opt$reference_sequence) # genome assembly
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
70 cat("done\n")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
71 outfile <- opt$output
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
72 # clean sequence names:
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
73 names(s) <- gsub(" .+", "", names(s))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
74 cat("reading gff...")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
75 g <- rtracklayer::import(opt$gff3, format = 'gff3') # DANTE gff3
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
76 cat("done\n")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
77 # testing
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
78 if (FALSE) {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
79 g <- rtracklayer::import("./test_data/sample_ltr_annotation.gff3")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
80 s <- readDNAStringSet("./test_data/sample_genome.fasta")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
81
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
82 g <- rtracklayer::import("./test_data/DANTE_LTR_Vfaba_chr5.gff3")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
83 s <- readDNAStringSet("./test_data/211010_Vfaba_chr5.fasta")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
84 names(s) <- gsub(" .+", "", names(s))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
85 ncpus <- 10
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
86 lineage_info <- read.table("databases/lineage_domain_order.csv", sep = "\t", header =
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
87 TRUE, as.is = TRUE)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
88 source("./R/ltr_utils.R")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
89 }
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
90
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
91
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
92 # get te sequence based on rank
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
93
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
94 # evaluate best TE - DLTP grou
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
95 s_te <- get_te_sequences(g, s) # split by 'element quality'
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
96 # best quality - split by lineage
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
97 word_size <- 15
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
98 # best TE
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
99 TE_DLTP_info <- analyze_TE(s_te$DLTP, word_size = word_size, ncpus = ncpus)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
100
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
101 # TE rank 2:
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
102 TE_DLT_plus_DLP_info <- analyze_TE(c(s_te$DLP, s_te$DLT), word_size = word_size, ncpus
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
103 = ncpus)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
104 TE_DLT_plus_DLP_info_DLTP_verified <- compare_TE_datasets(c(s_te$DLT, s_te$DLP), ncpus
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
105 = ncpus,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
106 TE_DLTP_info$seqs_representative,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
107 word_size = word_size
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
108 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
109
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
110 TE_DLT_plus_DLP_info_multiplicity <- verify_based_on_multiplicity(TE_DLT_plus_DLP_info)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
111
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
112 # create additional library from rank 2 verified by multiplicity
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
113 id_for_additional_library <- setdiff(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
114 TE_DLT_plus_DLP_info_multiplicity$id_ok_mp_verified,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
115 TE_DLT_plus_DLP_info_DLTP_verified$id_ok_verified)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
116
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
117 if (length(id_for_additional_library) > 1) {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
118 seqs_for_additional_library <- c(s_te$DLP, s_te$DLT)[names(c(s_te$DLP, s_te$DLT)) %in%
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
119 id_for_additional_library]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
120 seqs_additional_info <- analyze_TE(seqs_for_additional_library, word_size =
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
121 word_size, ncpus = ncpus)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
122 seq_representative <- c(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
123 TE_DLTP_info$seqs_representative,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
124 seqs_additional_info$seqs_representative
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
125 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
126 }else {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
127 if (length(id_for_additional_library) == 1) {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
128 seq_representative <- c(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
129 TE_DLTP_info$seqs_representative,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
130 c(s_te$DLP, s_te$DLT)[names(c(s_te$DLP, s_te$DLT)) %in% id_for_additional_library]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
131 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
132 }else {
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
133 seq_representative <- TE_DLTP_info$seqs_representative
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
134 }
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
135 }
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
136
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
137
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
138 # TE rank 3
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
139 TE_DL_info_DLTP_verified <- compare_TE_datasets(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
140 s_te$DL,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
141 TE_DLTP_info$seqs_representative, min_coverage = 0.98,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
142 ncpus = ncpus
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
143 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
144
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
145
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
146 R <- seq_diversity(seq_representative)$richness
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
147 SI <- seq_diversity(seq_representative)$shannon_index
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
148
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
149 # final RM library:
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
150 seq_representative_no_ssr <- seq_representative[R > 20]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
151
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
152 ID <- g$ID[g$type == "transposable_element"]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
153 names(ID) <- paste0(seqnames(g), "_",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
154 start(g), "_",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
155 end(g), "#",
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
156 g$Final_Classification
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
157 )[g$type %in% "transposable_element"]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
158
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
159 # create clean gff3
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
160 id_of_good_te <- unique(c(
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
161 TE_DLTP_info$te_conflict_info$ok,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
162 TE_DLT_plus_DLP_info_DLTP_verified$id_ok_verified,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
163 TE_DLT_plus_DLP_info_multiplicity$id_ok_mp_verified,
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
164 TE_DL_info_DLTP_verified$id_ok_verified)
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
165 )
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
166
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
167 c1 <- g$ID %in% ID[id_of_good_te]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
168 c2 <- sapply(g$Parent, function(x)ifelse(length(x) == 0, "", x)) %in% ID[id_of_good_te]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
169
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
170 gff_out <- g[c1 | c2]
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
171
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
172
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
173 writeXStringSet(seq_representative, paste0(opt$output, "_RM_lib.fasta"))
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
174 export(gff_out, paste0(opt$output, "_clean.gff3"), format = "gff3")
7b0bbe7477c4 "planemo upload commit 92c684dff3b377c8c08654c7f3d46a133385e3e0-dirty"
petr-novak
parents:
diff changeset
175