|
jpayne@69
|
1 #!/bin/bash
|
|
jpayne@69
|
2
|
|
jpayne@69
|
3 usage(){
|
|
jpayne@69
|
4 echo "
|
|
jpayne@69
|
5 Written by Brian Bushnell
|
|
jpayne@69
|
6 Last modified February 10, 2020
|
|
jpayne@69
|
7
|
|
jpayne@69
|
8 Description: Finds orfs and calls genes in unspliced prokaryotes.
|
|
jpayne@69
|
9 This includes bacteria, archaea, viruses, and mitochondria.
|
|
jpayne@69
|
10 Can also predict 16S, 23S, 5S, and tRNAs.
|
|
jpayne@69
|
11
|
|
jpayne@69
|
12 Usage: callgenes.sh in=contigs.fa out=calls.gff outa=aminos.faa out16S=16S.fa
|
|
jpayne@69
|
13
|
|
jpayne@69
|
14 File parameters:
|
|
jpayne@69
|
15 in=<file> A fasta file; the only required parameter.
|
|
jpayne@69
|
16 out=<file> Output gff file.
|
|
jpayne@69
|
17 outa=<file> Amino acid output.
|
|
jpayne@69
|
18 out16s=<file> 16S output.
|
|
jpayne@69
|
19 model=<file> A pgm file or comma-delimited list.
|
|
jpayne@69
|
20 If unspecified a default model will be used.
|
|
jpayne@69
|
21 stats=stderr Stats output (may be stderr, stdin, a file, or null).
|
|
jpayne@69
|
22 hist=null Gene length histogram.
|
|
jpayne@69
|
23 compareto= Optional reference gff file to compare with the gene calls.
|
|
jpayne@69
|
24 'auto' will name it based on the input file name.
|
|
jpayne@69
|
25
|
|
jpayne@69
|
26 Formatting parameters:
|
|
jpayne@69
|
27 json=false Print stats in JSON.
|
|
jpayne@69
|
28 binlen=20 Histogram bin length.
|
|
jpayne@69
|
29 bins=2000 Maximum histogram bins.
|
|
jpayne@69
|
30 pz=f (printzero) Print histogram lines with zero count.
|
|
jpayne@69
|
31
|
|
jpayne@69
|
32
|
|
jpayne@69
|
33
|
|
jpayne@69
|
34 Other parameters:
|
|
jpayne@69
|
35 minlen=60 Don't call genes shorter than this.
|
|
jpayne@69
|
36 trd=f (trimreaddescription) Set to true to trim read headers after
|
|
jpayne@69
|
37 the first whitespace. Necessary for IGV.
|
|
jpayne@69
|
38 merge=f For paired reads, merge before calling.
|
|
jpayne@69
|
39 detranslate=f Output canonical nucleotide sequences instead of amino acids.
|
|
jpayne@69
|
40 recode=f Re-encode nucleotide sequences over called genes, leaving
|
|
jpayne@69
|
41 non-coding regions unchanged.
|
|
jpayne@69
|
42
|
|
jpayne@69
|
43 Please contact Brian Bushnell at bbushnell@lbl.gov if you encounter any problems.
|
|
jpayne@69
|
44 "
|
|
jpayne@69
|
45 }
|
|
jpayne@69
|
46
|
|
jpayne@69
|
47 #This block allows symlinked shellscripts to correctly set classpath.
|
|
jpayne@69
|
48 pushd . > /dev/null
|
|
jpayne@69
|
49 DIR="${BASH_SOURCE[0]}"
|
|
jpayne@69
|
50 while [ -h "$DIR" ]; do
|
|
jpayne@69
|
51 cd "$(dirname "$DIR")"
|
|
jpayne@69
|
52 DIR="$(readlink "$(basename "$DIR")")"
|
|
jpayne@69
|
53 done
|
|
jpayne@69
|
54 cd "$(dirname "$DIR")"
|
|
jpayne@69
|
55 DIR="$(pwd)/"
|
|
jpayne@69
|
56 popd > /dev/null
|
|
jpayne@69
|
57
|
|
jpayne@69
|
58 #DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )/"
|
|
jpayne@69
|
59 CP="$DIR""current/"
|
|
jpayne@69
|
60
|
|
jpayne@69
|
61 z="-Xmx6g"
|
|
jpayne@69
|
62 z2="-Xms6g"
|
|
jpayne@69
|
63 set=0
|
|
jpayne@69
|
64
|
|
jpayne@69
|
65 if [ -z "$1" ] || [[ $1 == -h ]] || [[ $1 == --help ]]; then
|
|
jpayne@69
|
66 usage
|
|
jpayne@69
|
67 exit
|
|
jpayne@69
|
68 fi
|
|
jpayne@69
|
69
|
|
jpayne@69
|
70 calcXmx () {
|
|
jpayne@69
|
71 source "$DIR""/calcmem.sh"
|
|
jpayne@69
|
72 setEnvironment
|
|
jpayne@69
|
73 parseXmx "$@"
|
|
jpayne@69
|
74 }
|
|
jpayne@69
|
75 calcXmx "$@"
|
|
jpayne@69
|
76
|
|
jpayne@69
|
77 function callgenes() {
|
|
jpayne@69
|
78 local CMD="java $EA $EOOM $z $z2 -cp $CP prok.CallGenes $@"
|
|
jpayne@69
|
79 #Too long to echo sometimes since wildcards can be expanded
|
|
jpayne@69
|
80 #echo $CMD >&2
|
|
jpayne@69
|
81 eval $CMD
|
|
jpayne@69
|
82 }
|
|
jpayne@69
|
83
|
|
jpayne@69
|
84 callgenes "$@"
|
