Mercurial > repos > petr-novak > repeat_annotation_pipeline3

comparison clean_rm_output.R @ 0:ea6a3059a6af draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author petr-novak
date Mon, 18 Oct 2021 11:01:20 +0000
parents
children 814cba36e435
comparison
equal deleted inserted replaced
-1:000000000000 0:ea6a3059a6af
1 #!/usr/bin/env Rscript
2 suppressPackageStartupMessages(library(rtracklayer))
3
4 gff_cleanup = function(gff){
5 ## remove overlapin annotation track - assign new annot
6 gff_disjoin = disjoin(gff, with.revmap=TRUE)
7 ## append annotation:
8 gff_names = mclapply(as.list(gff_disjoin$revmap), FUN = function(x)gff$Name[x], mc.cores = 8)
9 gff_strands = mclapply(as.list(gff_disjoin$revmap), FUN = function(x)strand(gff[x]), mc.cores = 8)
10 new_annot = sapply(sapply(gff_names, unique), paste, collapse="|")
11 new_annot_uniq = unique(new_annot)
12 lca_annot = sapply(strsplit(new_annot_uniq, "|", fixed = TRUE), resolve_name)
13 names(lca_annot) = new_annot_uniq
14 new_annot_lca = lca_annot[new_annot]
15 #new_annot_lca = sapply(sapply(gff_names, unique), resolve_name)
16 strand_attribute = sapply(sapply(gff_strands, unique), paste, collapse="|")
17 gff_disjoin$strands=strand_attribute
18 gff_disjoin$source="RM"
19 gff_disjoin$type="repeat"
20 gff_disjoin$score=NA
21 gff_disjoin$phase=NA
22 gff_disjoin$Name=new_annot_lca
23 gff_disjoin$Original_names=new_annot
24 gff_disjoin$revmap=NULL
25 return(gff_disjoin)
26 }
27
28 resolve_name=function(x){
29 if (length(x)==1){
30 # no conflict
31 return(x)
32 } else{
33 y = sapply(x, strsplit, split="/", fixed = TRUE)
34 ny = table(unlist(sapply(y, function(x)paste(seq_along(x),x))))
35 if (max(ny)<length(x)){
36 return("Unknown")
37 }else{
38 k = which(ny==length(x))
39 r = max(as.numeric((gsub(" .+","",names(k)))))
40 out = paste(y[[1]][1:r], collapse="/")
41 return(out)
42 }
43 }
44 }
45
46
47
48 infile = commandArgs(T)[1]
49 outfile = commandArgs(T)[2]
50
51 ## infile = "./test_data/raw_rm.out"
52
53 rm_out = read.table(infile, as.is=TRUE, sep="", skip = 2, fill=TRUE, header=FALSE)
54
55 gff = GRanges(seqnames = rm_out$V5, ranges = IRanges(start = rm_out$V6, end=rm_out$V7))
56
57 # repeat class after # symbol - syntax 1
58 gff$Name=rm_out$V11
59
60 ## is repeat type is specifies by double underscore:
61 ## then rm_out$V11 is unspecified
62 if (any(rm_out$V11 == "Unspecified")){
63 ## set Name from prefix
64 ## TODO
65 inc = rm_out$V11 == "Unspecified"
66 Name = gsub("__.+","",rm_out$V10)
67 # chanche Usnpsecified to new name
68 gff$Name[inc] = Name
69 }
70
71
72 ## join neighbors with the same annotation, disregard strand!
73 result <- unlist(reduce(split(gff, gff$Name)))
74
75 result$Name <- names(result)
76
77 result_clean = gff_cleanup(result)
78
79 ## TODO
80 ## identify conflicting annotation, replace by LCA but keep origin list of classifications
81
82 gff_out = sortSeqlevels(result_clean)
83 gff_out = sort(gff_out)
84 gff_out$type = "repeat_region"
85 gff_out$source = "RepeatMasker_parsed"
86 gff_out$ID=paste0(gff_out$Name, "_", seq_along(gff_out$Name))
87 export(gff_out, format = "gff3", con=outfile)
88
89