annotate splicetrap/bin/mapping_rmap.sh @ 7:37a16ff93dd9 draft default tip

planemo upload
author bioitcore
date Thu, 12 Oct 2017 16:26:36 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
7
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
1 #/data/zhang/wuj/tools/bowtie-0.12.3/bowtie -a --best /data/zhang/wuj/database/hg18/AS/TXdb.2/btw/TXdb -f -t s_1_sequence.txtparta >s_1.map &
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
2 #SrcFolder='/data/zhang/wuj/scripts/SpliceTrap.0.8'
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
3 InputFileName=$1
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
4 faorfq=$2
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
5 DatabasePrefix=$3
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
6 Outputfolder=$4
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
7 SrcFolder=$5
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
8 DatabaseFolder=$SrcFolder'/../db/'$DatabasePrefix'/TXdb.fasta'
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
9 TmpFolderName=`basename $1`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
10
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
11 cd $Outputfolder;
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
12 #prepare the folder
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
13 if [ -d $TmpFolderName.result ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
14 echo "MAPPING: !!!Error, there is already a folder named "$TmpFolderName".result !"
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
15 echo "MAPPING: !!!change the name of that folder first in case I erase them..."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
16 exit
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
17 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
18 echo "MAPPING: Start to map $InputFileName....Creating cache folder $TmpFolderName.result"
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
19 mkdir $TmpFolderName".result"
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
20 cd $TmpFolderName".result"
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
21 mkdir cache stat
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
22 cd cache
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
23 echo "MAPPING: Split file..."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
24 split -l 1000000 $InputFileName
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
25 echo "MAPPING: generating shell scirpts...."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
26 for name in x*
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
27 do
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
28
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
29 echo "rmap -M 100 -m 2 -c $DatabaseFolder -o $name.rmap $name; perl $SrcFolder/rmap2eland.pl $name.rmap $name $name.eland;rm $name.rmap ;perl $SrcFolder/mark.mt.4eland.pl $name.eland >$name.nomt;rm $name.eland">>map.sh
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
30 echo $name >>checklist
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
31 done
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
32 tasknum=`wc -l map.sh |tr -d "\n"`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
33 perl $SrcFolder/batchqsub.pl map.sh
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
34
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
35 echo "MAPPING: map $InputFileName to TXdb done...start to check.."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
36 while [ 1 ]
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
37 do
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
38 if [ -f mapcheck.sh ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
39 rm mapcheck.sh
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
40 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
41 while read checklist
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
42 do
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
43
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
44 name=`echo $checklist |tr -d "\n"`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
45 echo "MAPPING: checking $name...."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
46 readnum=`wc -l $name | cut -f1 -d" "`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
47 if [ $faorfq == "fasta" ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
48 readnum=`echo "$readnum/2"|bc`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
49 else
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
50 readnum=`echo "$readnum/4"|bc`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
51 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
52 if [ -f $name.nomt ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
53 rmapnum=`wc -l $name.nomt | cut -f1 -d" "`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
54 else
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
55 rmapnum=0
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
56 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
57 if [ $rmapnum != $readnum ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
58 echo "rmap -M 100 -m 2 -c $DatabaseFolder -o $name.rmap $name; perl $SrcFolder/rmap2eland.pl $name.rmap $name $name.eland;rm $name.rmap ;perl $SrcFolder/mark.mt.4eland.pl $name.eland >$name.nomt;rm $name.eland">>mapcheck.sh
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
59
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
60 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
61 done <checklist
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
62 if [ -f mapcheck.sh ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
63 checktasknum=`wc -l mapcheck.sh |tr -d "\n"`
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
64 if [ $checktasknum == $tasknum ];then
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
65 echo "MAPPING: warning! none of the mapping tasks properly finished!"
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
66 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
67 echo "MAPPING: resubmiting TASKS...."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
68 perl $SrcFolder/batchqsub.pl mapcheck.sh
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
69 else
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
70 break
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
71 fi
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
72 done
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
73 echo "MAPPING: Mapping is really done....merging files..."
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
74
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
75
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
76 cat *.nomt >$Outputfolder/$TmpFolderName.nomt
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
77 cd ../../
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
78
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
79 rm $TmpFolderName.result -rf
37a16ff93dd9 planemo upload
bioitcore
parents:
diff changeset
80