ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/owl/trunk/modelit/scripts/runall.sh
Revision: 1299
Committed: Tue Jan 11 17:38:33 2011 UTC (9 years, 8 months ago) by hstehr
File size: 5567 byte(s)
Log Message:
adding Model-It server files to repository
Line User Rev File contents
1 hstehr 1299 #!/bin/sh
2     # Script to run the whole CASP prediction pipeline
3     # Cluster ready: to submit it with qsub, just make sure you use absolute paths
4    
5    
6     ## CONSTANTS
7     bindir=/project/StruPPi/CASP8/scripts
8     tmpdir=/tmp
9    
10     DEFAULT_PDBBLASTDB=seqs_pdbase_20080903.fix.reps.fa
11     DEFAULT_MAX_HITS=20
12     DEFAULT_PSIBLAST_ITER=5
13    
14     paulSpeed=slow
15    
16     cts="Ca,Cg"
17     dists="8.0,8.0"
18     numDistgeomModels=40
19     DEFAULT_CCT="0.4"
20    
21     DEFAULT_GTG_DIR="/project/StruPPi/CASP8/gtg"
22    
23     CASP_AUTHOR="3757_3251_6331"
24    
25     CASP_METHOD_FILE="/project/StruPPi/CASP8/scripts/method.txt"
26    
27     ## OPTIONS
28    
29     targetseqfile=""
30     outdir=""
31     pdbblastdb="$DEFAULT_PDBBLASTDB"
32     psiblastopt=""
33     psiblastIter="$DEFAULT_PSIBLAST_ITER"
34     templatefile=""
35     skiptinkeropt=""
36     CCT="$DEFAULT_CCT"
37     selectTemplates=""
38     gtgDir=$DEFAULT_GTG_DIR
39     maxHits=$DEFAULT_MAX_HITS
40     noPhiPsi=""
41     noTransOmega=""
42    
43     while getopts i:o:Kj:DP:G:t:s:l:x:FO opt
44     do
45     case "$opt" in
46     i) targetseqfile="$OPTARG";;
47     o) outdir="$OPTARG";;
48     K) psiblastopt="-K";;
49     j) psiblastIter="$OPTARG";;
50     D) skiptinkeropt="-D";;
51     P) pdbblastdb="$OPTARG";;
52     G) gtgDir="$OPTARG";;
53     t) templatefile="$OPTARG";;
54     s) CCT="$OPTARG";;
55     l) selectTemplates="$OPTARG";;
56     x) maxHits="$OPTARG";;
57     F) noPhiPsi="-F";;
58     O) noTransOmega="-O";;
59     esac
60     done
61    
62    
63     if [ -z "$targetseqfile" ] || [ -z "$outdir" ]
64     then
65     echo "Usage: $0 "
66     echo " -i <file> : the target sequence file"
67     echo " -o <dir> : the output directory"
68     echo " [-K] : no psiblast, just blast"
69     echo " [-j <number>] : number of psiblast iterations, default: $DEFAULT_PSIBLAST_ITER"
70     echo " [-D] : no Tinker reconstruction, just output average graph"
71     echo " [-P <blast db>] : PDB blast db, default: $DEFAULT_PDBBLASTDB"
72     echo " [-G <dir>] : directory containing GTG output, default: $DEFAULT_GTG_DIR"
73     echo " [-t <file>] : specifying a templates list file, no template selection step is performed (skip STEP 1)"
74     echo " [-s <number>] : CCT (contact conservation threshold) for the graph averaging step, default: $DEFAULT_CCT"
75     echo " [-l <B, P or G>]: forces selection of (B)LAST template list, (P)SIBLAST template list or (G)TG template list (even if they are empty!)"
76     echo " [-x <number>] : a hard maximum for the number of hits to be taken in the template selection step, default: $DEFAULT_MAX_HITS"
77     echo " [-F] : don't use phi/psi constraints for reconstruction. Default: phi/psi constraints used"
78     echo " [-O] : don't enforce trans conformation for omega torsion angle (peptide bond). Default: enforcing"
79     echo ""
80     echo "REMEMBER: always use absolute paths to run this script in the cluster"
81     echo ""
82     exit 1
83    
84     fi
85    
86    
87    
88    
89    
90     ## file definitions
91     basename=`basename $targetseqfile .fa`
92     templatesAlnFile=$outdir/$basename.templates_aln.fasta
93     target2templAlnFile=$outdir/$basename.target2templ.fasta
94     tcoffeelog=$outdir/$basename.tcoffee.log
95     ssCompareFile=$outdir/$basename.ss.compare
96     psipredFile=$outdir/$basename.horiz #input file!
97    
98     ### STEP 1: template selection
99     if [ -z "$templatefile" ]
100     then
101     echo "################################################"
102     echo "# STEP 1: TEMPLATE SELECTION"
103     echo "################################################"
104    
105     templatefile=$outdir/$basename.templates
106    
107     selectTemplatesOpt=""
108     if [ -n "$selectTemplates" ]
109     then
110     selectTemplatesOpt="-l $selectTemplates"
111     fi
112     $bindir/templateSelection -i $targetseqfile -o $outdir -P $pdbblastdb -b $basename -G $gtgDir $psiblastopt $selectTemplatesOpt -x $maxHits -j $psiblastIter
113    
114     numTemps=`cat $templatefile | wc -l`
115     if [ "$numTemps" -eq "0" ]
116     then
117     echo "NO TEMPLATES FOUND, CAN'T MODEL!"
118     exit 1
119     fi
120     fi
121    
122     ### STEP 2: structural alignment of templates
123     echo "################################################"
124     echo "# STEP 2: STRUCTURAL ALIGNMENT OF TEMPLATES"
125     echo "################################################"
126     $bindir/align_paul.sh $templatefile $templatesAlnFile $tmpdir $paulSpeed
127     if [ "$?" -ne "0" ]
128     then
129     echo "ERROR: paul alignment failed. Exiting"
130     exit 1
131     fi
132    
133    
134     ### STEP 3: target sequence to templates profile alignment
135     echo "################################################"
136     echo "# STEP 3: TARGET TO TEMPLATES ALIGNMENT"
137     echo "################################################"
138     $bindir/align_seq2prof.sh $targetseqfile $templatesAlnFile $target2templAlnFile $tcoffeelog
139     echo "Wrote target to template alignment to $target2templAlnFile"
140     echo "Wrote t-coffee output to $tcoffeelog"
141     # generate secondary structure comparison, needs psipred output file, searching in output dir
142     if [ -e $psipredFile ]
143     then
144     $bindir/displaySSMatching -n -a $target2templAlnFile -s $psipredFile > $ssCompareFile
145     echo "Wrote secondary structure comparison to $ssCompareFile"
146     else
147     echo "Could not find psipred output, please run displaySSMatching manually to view secondary structure comparison."
148     fi
149    
150     ### STEP 4: graph average and reconstruction
151     echo "################################################"
152     echo "# STEP 4: GRAPH AVERAGING AND RECONSTRUCTION"
153     echo "################################################"
154    
155    
156     if [ -z "$skiptinkeropt" ]
157     then
158     $bindir/averageGraph -f $targetseqfile -P $templatefile -t $cts -d $dists -b $basename -o $outdir -a $target2templAlnFile -s $CCT \
159     -r $numDistgeomModels -c $CASP_AUTHOR -m $CASP_METHOD_FILE $noPhiPsi $noTransOmega
160     else
161     $bindir/averageGraph -f $targetseqfile -P $templatefile -t $cts -d $dists -b $basename -o $outdir -a $target2templAlnFile -s $CCT
162     echo "Skipping reconstruction step. No 3D model will be written."
163     fi

Properties

Name Value
svn:executable *