comparison orthologs/ucsb_hamster/ucsb_galaxy_hamster.sh @ 0:5b9a38ec4a39 draft default tip

First commit of old repositories
author osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
date Tue, 11 Mar 2014 12:19:13 -0700
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:5b9a38ec4a39
1 #!/bin/sh
2 # set -x
3 # Uncomment 'set -x' for debug information
4
5 #############################################
6 # UCSB HAMSTER - GALAXY HISTORY #
7 # #
8 # Executed when user chooses Galaxy history #
9 #############################################
10
11 # Set your hamster script location here. The directory containing these scripts should be in
12 #your path
13 script="hamstrsearch_local-hmmer3.pl"
14 # Set your unbuild.py script location here
15 unbuild="unbuild.py"
16 # Set your emap2fasta.pl script location here
17 emap2fasta="emap2fasta.pl"
18
19
20 # 1 - Sequence input file
21 # 2 - Proteins results output file
22 # 3 - CDS results output file
23 # 4 - Screen log
24 # 5 - Species name
25 # 6 - whether to use EST flag D=DNA so use -est flag P=Protein so do not use -est flag in hmmstr call
26 # 7 - HMM Input from UCSB HMMBUILD
27 # 8 - MUSCLE data from UCSB MUSCLE
28 # 9 - Reference Species File
29 # 10 - Reference Species Name
30
31 input=$1
32 proteins=$2
33 cdsfile=$3
34 screenlog=$4
35 speciesName=$5
36 datatype=$6
37 hmm_data=$7
38 muscle_data=$8
39 filepath=`pwd`
40 tail="_prot"
41 tail2="_temp"
42
43 # set flag based on input
44 if [ $datatype = "P" ];
45 then
46 estflag="-protein"
47 else
48 estflag="-est"
49 fi
50
51 refspfile=${9}
52 refsphist=${10}
53
54
55
56 echo "Protein or EST? : $estflag" >> $screenlog
57 echo "Reference genome file from galaxy history: $refspfile" >> $screenlog
58 echo "Reference species genome name: $refsphist" >> $screenlog
59
60 # unbuild.py here on $hmm_data
61 mkdir core
62 mkdir core/hmm_dir
63 cp $hmm_data core/core.fa
64
65 $unbuild core/hmm_dir core/core.fa
66 cp core/hmm_dir/hmmlist.txt core/hmmlist.txt
67
68 # use formatdb to generate new blastdb from this input file
69 refsphistGALAXY=$refsphist
70 mkdir $refsphistGALAXY
71
72 cp $muscle_data $refsphistGALAXY/$refsphist$tail2
73 $emap2fasta $refsphistGALAXY/$refsphist$tail2 $refsphist
74 cp full.fasta core/core.fa
75
76 cp $refspfile $refsphistGALAXY/$refsphist$tail
77 cd $refsphistGALAXY
78 formatdb -t $refsphist -i $refsphist$tail -n $refsphist$tail
79
80 echo "*** Direcotry Structure of Ref. Genome ***" >> $screenlog
81 ls -l >> $screenlog
82 echo >> $screenlog
83
84 cd $filepath
85
86 # script execution
87 $script -sequence_file=$1 $estflag -taxon=$5 -hmmset=core -refspec=$refsphistGALAXY -galaxyout=$2 -2galaxyout=$cdsfile 2>log.txt >> $screenlog