Mercurial > repos > koen > gbsx

comparison gbsx_digest.sh @ 22:062399f3787f draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
new version of gbsx: v1.2 with barcodeGenerator, demultiplex and digest
author koen
date Tue, 26 Jan 2016 07:38:36 -0500
parents
children
comparison
equal deleted inserted replaced
21:32ee036c156a 22:062399f3787f
1 #!/usr/bin/env bash
2
3 #export PATH=$PATH:$(dirname $0)
4 #BASEDIR=$(dirname $0)
5
6
7 genome=$1
8 length=$2
9 enzyme_combination=$3
10 dualEnzyme="0"
11 min_size=$4
12 max_size=$5
13 digest_bed=$6
14 digest_result=$7
15
16 awk 'BEGIN {n_seq=0;} /^>/ {if(n_seq%1==0){file=$1".fa"; sub(">","",file)} print >> file; n_seq++; print file >> "list.txt"; next;} { print >> file; }' < $genome
17
18 case "$enzyme_combination" in
19 ApeKI)
20 enzyme="ApeKI"
21 cutsite="G^CWGC"
22 ;;
23 PstI)
24 enzyme="PstI"
25 cutsite="CTGCA^G"
26 ;;
27 EcoT22I)
28 enzyme="EcoT22I"
29 cutsite="ATGCA^T"
30 ;;
31 PasI)
32 enzyme="PasI"
33 cutsite="CC^CWGGG"
34 ;;
35 HpaII)
36 enzyme="HpaII"
37 cutsite="C^CGG"
38 ;;
39 MspI)
40 enzyme="MspI"
41 cutsite="C^CGG"
42 ;;
43 PstI-EcoT22I)
44 dualEnzyme="1"
45 enzyme="PstI"
46 cutsite="CTGCA^G"
47 enzyme2="EcoT22I"
48 cutsite2="ATGCA^T";
49 ;;
50 PstI-MspI)
51 dualEnzyme="1"
52 enzyme="PstI"
53 cutsite="CTGCA^G"
54 enzyme2="MspI"
55 cutsite2="C^CGG"
56 ;;
57 PstI-TaqI)
58 dualEnzyme="1"
59 enzyme="PstI"
60 cutsite="CTGCA^G"
61 enzyme2="TaqI"
62 cutsite2="T^CGA"
63 ;;
64 SbfI-MspI)
65 dualEnzyme="1"
66 enzyme="SbfI"
67 cutsite="CCTGCA^GG"
68 enzyme2="MspI"
69 cutsite2="C^CGG"
70 ;;
71 AsiSI-MspI)
72 dualEnzyme="1"
73 enzyme="AsiSI"
74 cutsite="GCCGAT^CGC"
75 enzyme2="MspI"
76 cutsite2="C^CGG"
77 ;;
78 BssHII-MspI)
79 dualEnzyme="1"
80 enzyme="BssHII"
81 cutsite="G^CGCGC"
82 enzyme2="MspI"
83 cutsite2="C^CGG"
84 ;;
85 FseI-MspI)
86 dualEnzyme="1"
87 enzyme="FseI"
88 cutsite="GGCCGG^CC"
89 enzyme2="MspI"
90 cutsite2="C^CGG"
91 ;;
92 SalI-MspI)
93 dualEnzyme="1"
94 enzyme="SalI"
95 cutsite="G^TCGAC"
96 enzyme2="MspI"
97 cutsite2="C^CGG"
98 ;;
99 ApoI)
100 enzyme="ApoI"
101 cutsite="R^AATTY"
102 ;;
103 BamHI)
104 enzyme="BamHI"
105 cutsite="G^GATCC"
106 ;;
107 MseI)
108 enzyme="MseI"
109 cutsite="T^TAA"
110 ;;
111 Sau3AI)
112 enzyme="Sau3AI"
113 cutsite="^GATC"
114 ;;
115 RBSTA)
116 enzyme="RBSTA"
117 cutsite="^TA"
118 ;;
119 RBSCG)
120 enzyme="RBSCG"
121 cutsite="^CG"
122 ;;
123 NspI)
124 enzyme="NspI"
125 cutsite="RCATG^Y"
126 ;;
127 AvaII)
128 enzyme="AvaII"
129 cutsite="G^GWCC"
130 ;;
131 esac
132
133 extraparameters="";
134 if [ "$dualEnzyme" != "0" ];
135 then
136 extraparameters="-E $enzyme2 -D $cutsite2";
137 fi
138
139 perl $GBSX_PATH/GBSX_digest.pl -f list.txt -l $length -d $cutsite -e $enzyme $extraparameters -n $min_size -m $max_size;
140
141 rm *.fa;
142 rm list.txt;
143
144 if [ "$dualEnzyme" != "0" ];
145 then
146 cat "genome."$enzyme"."$enzyme2"."$length"nt.digest.bed" > $digest_bed;
147 cat "genome."$enzyme"."$enzyme2"."$length"nt.digest_results" > $digest_result;
148 else
149 cat "genome."$enzyme"."$length"nt.digest.bed" > $digest_bed;
150 cat "genome."$enzyme"."$length"nt.digest_results" > $digest_result;
151 fi
152
153