annotate Dotplot_Release/dotplot.bash @ 4:a0b4bcf67972 draft

Uploaded

1 #!/bin/bash
2 #SCRIPT=$(readlink -e $0)
3 #SCRIPTPATH=`dirname $SCRIPT`
4 pushd `dirname $0` > /dev/null
5 SCRIPTPATH=`pwd`
6 popd > /dev/null
7 
8 usage() { printf "Usage: $0 
9 [-f <saint_file_name.txt>]
10 [-i <0 for SaintExpress format, 1 for other>]
11 [-c <clustering to perform. Options: b (biclustering), h (hierarchical), n (none, requires input text files for bait and prey ordering; see options -b and -p)>]
12 [-n <clustering type to be performed if option -c is set to \"h\">]
13 [-d <distance metric to use if option -c is set to \"h\">]
14 [-b <list of bait proteins in display order (see option -c n)>]
15 [-p <list of prey proteins in display order (see option -c n). Set this to \"all\" if you want to include all preys and cluster them>]
16 [-s <primary FDR cutoff [0-1, recommended=0.01]>]
17 [-t <secondary FDR cutoff [must be less than the primary, recommended=0.025]>
18 [-x <spectral count minimum. Only preys with >= this will be used]>
19 [-m <maximum spectral count>]
20 [-N <normalization, 0 for no (default), 1 for yes, 2 for normalization based on significant preys counts (prey FDR <= option -t)>]
21 [-C <FDR cutoff for normalization if using option -N 2 (deafult is -t)>]\n"
22 1>&2; exit 1; }
23 
24 N=0
25 n="ward"
26 d="canberra"
27 x=0
28 i=0
29 while getopts ":f:i:s:t:x:m:c:n:d:b:p:N:C:" o; do
30     case "${o}" in
31         f)
32             f=${OPTARG}
33             ;;
34         i)
35 	    i=${OPTARG}
36             ;;
37         s)
38             s=${OPTARG}
39             ;;
40 	t)
41             t=${OPTARG}
42             ;;
43         x)
44 	    x=${OPTARG}
45             ;;
46 	m)
47             m=${OPTARG}
48             ;;
49 	c)
50             c=${OPTARG}
51 	    ;;
52 	n)
53 	    n=${OPTARG}
54 	    ;;
55 	d)
56 	    d=${OPTARG}
57 	    ;;
58 	b)
59             b=${OPTARG}
60 	    ;;
61 	p)
62 	    p=${OPTARG}
63 	    ;;
64 	N)
65 	    N=${OPTARG}
66 	    ;;
67 	C)
68 	    C=${OPTARG}
69 	    ;;
70         *)
71             usage
72             ;;
73     esac
74 done
75 shift $((OPTIND-1))
76 
77 filename=${f%%.*}
78 echo "Saint input file = ${f}"
79 echo "Primary FDR cutoff = ${s}"
80 echo "Secondary FDR cutoff for dotplot = ${t}"
81 echo "Minimum spectral count for significant preys = ${x}"
82 echo "Maximum spectral count for dot plot = ${m}"
83 
84 if [ -z "${f}" ] || [ -z "${s}" ] || [ -z "${t}" ] || [ -z "${m}" ] || [ -z "${c}" ]; then
85     usage
86 fi
87 
88 if [ "${i}" == 1 ]; then
89 	$SCRIPTPATH/SaintConvert.pl -i ${f}
90 	f="mockSaintExpress.txt"
91 fi
92 
93 if [ "${x}" -ge "${m}" ]; then
94 	echo "spectral count minimum (${x}) cannot be greater than or equal to the maximum (${m})"
95 	exit 1;
96 elif [ "${x}" -lt 0 ]; then
97 	echo "spectral count minimum (${x}) cannot be less than 0. Setting to 0 and continuing"
98 	x=0
99 fi
00 
01 ###Check for normalization
02 
03 if [ "${N}" == 1 ]; then
04 	printf "\nNormalization is being performed\n"
05 	$SCRIPTPATH/Normalization.R ${f}
06 	f="norm_saint.txt"
07 elif [ "${N}" == 2 ]; then
08 	printf "\nNormalization is being performed\n"
09 	if [ -z "${C}" ]; then
10 		C=${t}
11 	fi
12 	$SCRIPTPATH/Normalization_sigpreys.R ${f} ${C}
13 	f="norm_saint.txt"
14 fi
15 
16 
17 ###Check for clustering etc
18 
19 if [ "${c}" == "h" ] && [ -z "${n}" ]; then
20 	printf "\nHierarchial clustering was selected (-c = h), but no clustering method (-n) was chosen.\n"
21 	printf "The input parameter -n must be set to one of \"average\", \"centroid\", \"complete\", \"mcquitty\",\n"
22 	printf "\"median\", \"single\" or \"ward\". \"ward\" will be selected as default.\n\n"
23 	n="ward"
24 elif [ "${c}" == "h" ] && [ -n "${n}" ]; then
25 	if [ "${n}" == "average" ] || [ "${n}" == "centroid" ] || [ "${n}" == "complete" ] || [ "${n}" == "mcquitty" ] || [ "${n}" == "median" ] || [  "${n}" == "single" ] || [ "${n}" == "ward" ]; then
26 		printf "\nHierarchical clustering (method = ${n}) will be performed\n\n"
27 	else
28 		printf "\n${n} is not a valid Hierarchical clustering method.\n"
29 		printf "Choose one of \"average\", \"centroid\", \"complete\", \"mcquitty\", \"median\", \"single\" or \"ward\"\n\n"
30 		exit 1
31 	fi
32 fi
33 
34 p_c=0
35 if [ "${c}" == "h" ] && [ -z "${d}" ]; then
36 	printf "\nHierarchial clustering was selected (-c = h), but no distance metric (-d) was chosen.\n"
37 	printf "The input parameter -d must be set to one of  \"binary\", \"canberra\", \"euclidean\",\n"
38 	printf "\"manhattan\", \"maximum\" or \"minkowski\". \"canberra\" will be selected as default.\n\n"
39 	d="canberra"
40 elif [ "${c}" == "h" ] && [ -n "${d}" ]; then
41 	if [ "${d}" == "binary" ] || [ "${d}" == "canberra" ] || [ "${d}" == "euclidean" ] || [ "${d}" == "manhattan" ] || [ "${d}" == "maximum" ] || [  "${d}" == "minkowski" ]; then
42 		printf "\nHierarchical clustering (distance metric = ${d}) will be performed\n\n"
43 	else
44 		printf "\n${d} is not a valid Hierarchical clustering distance metric.\n"
45 		printf "Choose one of  \"binary\", \"canberra\", \"euclidean\", \"manhattan\", \"maximum\" or \"minkowski\"\n\n"
46 		exit 1
47 	fi
48 fi
49 
50 if [ "${c}" == "n" ] && [ -z "${b}" ]; then
51 	printf "\n\"No Clustering\" option was selected (-c = n), but no bait list was included (option -b).\n"
52 	printf "Bait list must be in .txt formart.\n\n"
53 	exit 1
54 elif [ "${c}" == "n" ] && [ -z "${p}" ]; then
55 	printf "\n\"No Clustering\" option was selected (-c = n), but no prey list was included (option -p).\n"
56 	printf "Prey list must be in .txt formart.\n\n"
57 	exit 1
58 elif [ "${c}" == "n" ] && [ "${p}" == "all" ]; then
59 	printf "\n\"No Clustering\" option was selected (-c = n) for baits, but preys will still be clustered.\n"
60 	printf "using \"ward\" and \"canberra\" as defaults or options as supplied on command line.\n\n"
61 	p="empty"
62 	p_c=1
63 	n="ward"
64 	d="canberra"
65 fi
66 
67 
68 ###Check number of baits
69 
70 bait_n=$(perl $SCRIPTPATH/BaitCheck.pl -i ${f})
71 echo "Number of baits = "$bait_n
72 printf "\n\n"
73 
74 if [ "${c}" == "b" ] && [ $bait_n == 2 ]; then
75 	printf "\nWarning only 2 baits are present. Biclustering will not performed.\n"
76 	printf "Hierarchical clustering (method = ward) will be performed instead.\n\n"
77 	c="h"
78 	n="ward"
79 fi
80 
81 
82 ###Generate plots
83 
84 if [ "${c}" == "b" ]; then
85 	printf "\nBiclustering will be performed\n\n"
86 	$SCRIPTPATH/Step1_data_reformating.R ${f} ${s} ${filename}
87 	$SCRIPTPATH/Step2_data_filtering.R ${filename}_matrix.txt ${x} ${filename}
88 	GSL_RNG_SEED=123  $SCRIPTPATH/Step3_nestedcluster ${filename}.dat $SCRIPTPATH/biclust_param.txt
89 	$SCRIPTPATH/Step4_biclustering.R ${filename}.dat
90 
91 	$SCRIPTPATH/SOFD.pl -i ${f} -s ${s} -x ${x}
92 	$SCRIPTPATH/R_dotPlot.R ${s} ${t} ${m}
93 	mkdir Output_${filename}
94 	mkdir Output_${filename}/TempData_${filename}
95 	mv bait_lists Output_${filename}/TempData_${filename}
96 	mv Clusters Output_${filename}/TempData_${filename}
97 	mv MCMCparameters Output_${filename}/TempData_${filename}
98 	mv NestedClusters Output_${filename}/TempData_${filename}
99 	mv NestedMu Output_${filename}/TempData_${filename}
00 	mv NestedSigma2 Output_${filename}/TempData_${filename}
01 	mv OPTclusters Output_${filename}/TempData_${filename}
02 	mv ${filename}_matrix.txt Output_${filename}/TempData_${filename}
03 	mv ${filename}.dat Output_${filename}/TempData_${filename}
04 	mv SC_data.txt Output_${filename}/TempData_${filename}
05 	mv FDR_data.txt Output_${filename}/TempData_${filename}
06 	mv clustered_matrix.txt Output_${filename}/TempData_${filename}
07 	mv singletons.txt Output_${filename}/TempData_${filename}
08 	mv bait2bait_matrix.txt Output_${filename}/TempData_${filename}
09 	mv baitClusters Output_${filename}/TempData_${filename}
10 	mv clusteredData Output_${filename}/TempData_${filename}
11 	mv dotplot.pdf Output_${filename}
12 	mv bait2bait.pdf Output_${filename} 
13 	mv estimated.pdf Output_${filename} 
14 	mv stats.pdf Output_${filename}
15 	cp $SCRIPTPATH/legend.pdf Output_${filename}
16 elif [ "${c}" == "h" ]; then
17 
18 	$SCRIPTPATH/SOFD.pl -i ${f} -s ${s} -x ${x}
19 	$SCRIPTPATH/R_dotPlot_hc.R ${s} ${t} ${m} ${n} ${d} $SCRIPTPATH
20 
21 	mkdir Output_${filename}
22 	mkdir Output_${filename}/TempData_${filename}
23 	mv dotplot.pdf Output_${filename}
24 	mv heatmap_borders.pdf Output_${filename}
25 	mv heatmap_no_borders.pdf Output_${filename}
26 	mv bait2bait.pdf Output_${filename}
27 	mv SC_data.txt Output_${filename}/TempData_${filename}
28 	mv FDR_data.txt Output_${filename}/TempData_${filename}
29 	cp $SCRIPTPATH/legend.pdf Output_${filename}
30 elif [ "${c}" == "n" ]; then
31 	
32 	$SCRIPTPATH/SOFD.pl -i ${f} -s ${s} -x ${x}
33 	echo "$SCRIPTPATH/R_dotPlot_nc.R ${s} ${t} ${m} ${b} $p_c ${p} ${n} ${d} $SCRIPTPATH"
34 	$SCRIPTPATH/R_dotPlot_nc.R ${s} ${t} ${m} ${b} $p_c ${p} ${n} ${d} $SCRIPTPATH
35 
36 	mkdir Output_${filename}
37 	mkdir Output_${filename}/TempData_${filename}
38 	mv dotplot.pdf Output_${filename}
39 	mv heatmap_borders.pdf Output_${filename}
40 	mv heatmap_no_borders.pdf Output_${filename}
41 	mv SC_data.txt Output_${filename}/TempData_${filename}
42 	mv FDR_data.txt Output_${filename}/TempData_${filename}
43 	cp $SCRIPTPATH/legend.pdf Output_${filename}
44 else
45 	printf -- "-c must be one of [b, h, n]:  b (biclustering), h (hierarchical), n (none, requires input text files for bait and prey ordering>\n"
46 	exit 1;
47 fi
48 
49 if [ "${N}" == "1" ] || [ "${N}" == "2" ]; then
50 	mv norm_saint.txt Output_${filename}/TempData_${filename}
51 fi
52