Mercurial > repos > pjbriggs > pal_finder

comparison pal_finder_wrapper.sh @ 0:3f908e7fff4f draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded first version to toolshed.
author pjbriggs
date Thu, 11 Dec 2014 09:23:24 -0500
parents
children 771ebe02636f
comparison
equal deleted inserted replaced
-1:000000000000 0:3f908e7fff4f
1 #!/bin/sh
2 #
3 # pal_finder_wrapper.sh: run pal_finder perl script as a Galaxy tool
4 #
5 # Usage: run_palfinder.sh FASTQ_R1 FASTQ_R2 MICROSAT_SUMMARY PAL_SUMMARY FILTERED_MICROSATS [OPTIONS]
6 # run_palfinder.sh --454 FASTA MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]
7 #
8 # Options:
9 #
10 # --primer-prefix PREFIX: prefix added to the beginning of all primer names (prPrefixName)
11 # --2merMinReps N: miniumum number of 2-mer repeat units to detect (0=ignore units of this size)
12 # --3merMinReps N
13 # --4merMinReps N
14 # --5merMinReps N
15 # --6merMinReps N
16 # --primer-mispriming-library FASTA: specify a Fasta file with sequences to avoid amplifying
17 # --primer-opt-size VALUE: optimum primer length
18 # --primer-min-size VALUE: minimum acceptable primer length
19 # --primer-max-size VALUE: maximum acceptable primer length
20 # --primer-min-gc VALUE: minimum allowable percentage of Gs and Cs in any primer
21 # --primer-max-gc VALUE: maximum allowable percentage of Gs and Cs
22 # --primer-gc-clamp VALUE: number of consecutive Gs and Cs at 3' end of both left and right primer
23 # --primer-max-end-gc VALUE: max number of Gs or Cs in last five 3' bases of left or right primer
24 # --primer-min-tm VALUE: minimum acceptable melting temperature (Celsius) for a primer oligo
25 # --primer-max-tm VALUE: maximum acceptable melting temperature (Celsius)
26 # --primer-opt-tm VALUE: optimum melting temperature (Celsius)
27 # --primer-pair-max-diff-tm VALUE: max difference between melting temps of left & right primers
28 # --output_config_file FNAME: write a copy of the config.txt file to FNAME
29 # --filter_microsats FNAME: run Graeme Fox's Perl script to filter and sort the
30 # microsatellites from pal_finder and write to FNAME
31 #
32 # pal_finder is available from http://sourceforge.net/projects/palfinder/
33 #
34 # primer3 is available from http://primer3.sourceforge.net/releases.php
35 # (nb needs version 2.0.0-alpha)
36 #
37 # Explicitly set the locations of the pal_finder script, data files and the primer3
38 # executable by setting the following variables in the environment:
39 #
40 # * PALFINDER_SCRIPT_DIR: location of the pal_finder Perl script (defaults to
41 # /usr/bin)
42 # * PALFINDER_DATA_DIR: location of the pal_finder data files (specifically
43 # config.txt and simple.ref; defaults to /usr/share/pal_finder_v0.02.04)
44 # * PRIMER3_CORE_EXE: name of the primer3_core program, which should include the
45 # full path if it's not on the Galaxy user's PATH (defaults to primer3_core)
46 #
47 echo $*
48 #
49 # Initialise locations of scripts, data and executables
50 #
51 # Set these in the environment to overide at execution time
52 : ${PALFINDER_SCRIPT_DIR:=/usr/bin}
53 : ${PALFINDER_DATA_DIR:=/usr/share/pal_finder_v0.02.04}
54 : ${PRIMER3_CORE_EXE:=primer3_core}
55 #
56 # Filter script is in the same directory as this script
57 PALFINDER_FILTER_PL=$(dirname $0)/pal_finder_filter.pl
58 if [ ! -f $PALFINDER_FILTER_PL ] ; then
59 echo No pal_finder_filter.pl script >&2
60 exit 1
61 fi
62 #
63 # Check that we have all the components
64 function have_program() {
65 local program=$1
66 local got_program=$(which $program 2>&1 | grep "no $(basename $program) in")
67 if [ -z "$got_program" ] ; then
68 echo yes
69 else
70 echo no
71 fi
72 }
73 if [ "$(have_program $PRIMER3_CORE_EXE)" == "no" ] ; then
74 echo "ERROR primer3_core missing: ${PRIMER3_CORE_EXE} not found" >&2
75 exit 1
76 fi
77 if [ ! -f "${PALFINDER_DATA_DIR}/config.txt" ] ; then
78 echo "ERROR pal_finder config.txt not found in ${PALFINDER_DATA_DIR}" >&2
79 exit 1
80 fi
81 if [ ! -f "${PALFINDER_SCRIPT_DIR}/pal_finder_v0.02.04.pl" ] ; then
82 echo "ERROR pal_finder_v0.02.04.pl not found in ${PALFINDER_SCRIPT_DIR}" >&2
83 exit 1
84 fi
85 #
86 # Initialise parameters used in the config.txt file
87 PRIMER_PREFIX="test"
88 MIN_2_MER_REPS=6
89 MIN_3_MER_REPS=0
90 MIN_4_MER_REPS=0
91 MIN_5_MER_REPS=0
92 MIN_6_MER_REPS=0
93 PRIMER_MISPRIMING_LIBRARY=$PALFINDER_DATA_DIR/simple.ref
94 PRIMER_OPT_SIZE=
95 PRIMER_MAX_SIZE=
96 PRIMER_MIN_SIZE=
97 PRIMER_MAX_GC=
98 PRIMER_MIN_GC=
99 PRIMER_GC_CLAMP=
100 PRIMER_MAX_END_GC=
101 PRIMER_OPT_TM=
102 PRIMER_MAX_TM=
103 PRIMER_MIN_TM=
104 PRIMER_PAIR_MAX_DIFF_TM=
105 OUTPUT_CONFIG_FILE=
106 FILTERED_MICROSATS=
107 #
108 # Collect command line arguments
109 if [ $# -lt 2 ] ; then
110 echo "Usage: $0 FASTQ_R1 FASTQ_R2 MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]"
111 echo " $0 --454 FASTA MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]"
112 exit
113 fi
114 if [ "$1" == "--454" ] ; then
115 PLATFORM="454"
116 FNA=$2
117 else
118 PLATFORM="Illumina"
119 FASTQ_R1=$1
120 FASTQ_R2=$2
121 fi
122 MICROSAT_SUMMARY=$3
123 PAL_SUMMARY=$4
124 shift; shift; shift; shift
125 #
126 # Collect command line options
127 while [ ! -z "$1" ] ; do
128 case "$1" in
129 --primer-prefix)
130 shift
131 PRIMER_PREFIX=$1
132 ;;
133 --2merMinReps)
134 shift
135 MIN_2_MER_REPS=$1
136 ;;
137 --3merMinReps)
138 shift
139 MIN_3_MER_REPS=$1
140 ;;
141 --4merMinReps)
142 shift
143 MIN_4_MER_REPS=$1
144 ;;
145 --5merMinReps)
146 shift
147 MIN_5_MER_REPS=$1
148 ;;
149 --6merMinReps)
150 shift
151 MIN_6_MER_REPS=$1
152 ;;
153 --primer-mispriming-library)
154 shift
155 PRIMER_MISPRIMING_LIBRARY=$1
156 ;;
157 --primer-opt-size)
158 shift
159 PRIMER_OPT_SIZE=$1
160 ;;
161 --primer-max-size)
162 shift
163 PRIMER_MAX_SIZE=$1
164 ;;
165 --primer-min-size)
166 shift
167 PRIMER_MIN_SIZE=$1
168 ;;
169 --primer-max-gc)
170 shift
171 PRIMER_MAX_GC=$1
172 ;;
173 --primer-min-gc)
174 shift
175 PRIMER_MIN_GC=$1
176 ;;
177 --primer-gc-clamp)
178 shift
179 PRIMER_GC_CLAMP=$1
180 ;;
181 --primer-max-end-gc)
182 shift
183 PRIMER_MAX_END_GC=$1
184 ;;
185 --primer-opt-tm)
186 shift
187 PRIMER_OPT_TM=$1
188 ;;
189 --primer-max-tm)
190 shift
191 PRIMER_MAX_TM=$1
192 ;;
193 --primer-min-tm)
194 shift
195 PRIMER_MIN_TM=$1
196 ;;
197 --primer-pair-max-diff-tm)
198 shift
199 PRIMER_PAIR_MAX_DIFF_TM=$1
200 ;;
201 --output_config_file)
202 shift
203 OUTPUT_CONFIG_FILE=$1
204 ;;
205 --filter_microsats)
206 shift
207 FILTERED_MICROSATS=$1
208 ;;
209 *)
210 echo Unknown option: $1 >&2
211 exit 1
212 ;;
213 esac
214 shift
215 done
216 #
217 # Check that primer3_core is available
218 got_primer3=`which $PRIMER3_CORE_EXE 2>&1 | grep -v "no primer3_core in"`
219 if [ -z "$got_primer3" ] ; then
220 echo ERROR primer3_core not found >&2
221 exit 1
222 fi
223 #
224 # Set up the working dir
225 if [ "$PLATFORM" == "Illumina" ] ; then
226 # Paired end Illumina data as input
227 if [ $FASTQ_R1 == $FASTQ_R2 ] ; then
228 echo ERROR R1 and R2 fastqs are the same file >&2
229 exit 1
230 fi
231 ln -s $FASTQ_R1
232 ln -s $FASTQ_R2
233 fastq_r1=$(basename $FASTQ_R1)
234 fastq_r2=$(basename $FASTQ_R2)
235 else
236 # 454 data as input
237 ln -s $FNA
238 fna=$(basename $FNA)
239 fi
240 ln -s $PRIMER_MISPRIMING_LIBRARY
241 PRIMER_MISPRIMING_LIBRARY=$(basename $PRIMER_MISPRIMING_LIBRARY)
242 mkdir Output
243 #
244 # Copy in the default config.txt file
245 /bin/cp $PALFINDER_DATA_DIR/config.txt .
246 #
247 # Update the config.txt file with new values
248 function set_config_value() {
249 local key=$1
250 local value=$2
251 local config_txt=$3
252 if [ -z "$value" ] ; then
253 echo "No value for $key, left as default"
254 else
255 echo Setting "$key" to "$value"
256 sed -i 's,^'"$key"' .*,'"$key"' '"$value"',' $config_txt
257 fi
258 }
259 # Input files
260 set_config_value platform $PLATFORM config.txt
261 if [ "$PLATFORM" == "Illumina" ] ; then
262 set_config_value inputFormat fastq config.txt
263 set_config_value pairedEnd 1 config.txt
264 set_config_value inputReadFile $fastq_r1 config.txt
265 set_config_value pairedReadFile $fastq_r2 config.txt
266 else
267 set_config_value inputFormat fasta config.txt
268 set_config_value pairedEnd 0 config.txt
269 set_config_value input454reads $fna config.txt
270 fi
271 # Output files
272 set_config_value MicrosatSumOut Output/microsat_summary.txt config.txt
273 set_config_value PALsummaryOut Output/PAL_summary.txt config.txt
274 # Microsat info
275 set_config_value 2merMinReps $MIN_2_MER_REPS config.txt
276 set_config_value 3merMinReps $MIN_3_MER_REPS config.txt
277 set_config_value 4merMinReps $MIN_4_MER_REPS config.txt
278 set_config_value 5merMinReps $MIN_5_MER_REPS config.txt
279 set_config_value 6merMinReps $MIN_6_MER_REPS config.txt
280 # Primer3 settings
281 set_config_value primer3input Output/pr3in.txt config.txt
282 set_config_value primer3output Output/pr3out.txt config.txt
283 set_config_value primer3executable $PRIMER3_CORE_EXE config.txt
284 set_config_value prNamePrefix ${PRIMER_PREFIX}_ config.txt
285 set_config_value PRIMER_MISPRIMING_LIBRARY "$PRIMER_MISPRIMING_LIBRARY" config.txt
286 set_config_value PRIMER_OPT_SIZE "$PRIMER_OPT_SIZE" config.txt
287 set_config_value PRIMER_MIN_SIZE "$PRIMER_MIN_SIZE" config.txt
288 set_config_value PRIMER_MAX_SIZE "$PRIMER_MAX_SIZE" config.txt
289 set_config_value PRIMER_MIN_GC "$PRIMER_MIN_GC" config.txt
290 set_config_value PRIMER_MAX_GC "$PRIMER_MAX_GC" config.txt
291 set_config_value PRIMER_GC_CLAMP "$PRIMER_GC_CLAMP" config.txt
292 set_config_value PRIMER_MAX_END_GC "$PRIMER_MAX_END_GC" config.txt
293 set_config_value PRIMER_MIN_TM "$PRIMER_MIN_TM" config.txt
294 set_config_value PRIMER_MAX_TM "$PRIMER_MAX_TM" config.txt
295 set_config_value PRIMER_OPT_TM "$PRIMER_OPT_TM" config.txt
296 set_config_value PRIMER_PAIR_MAX_DIFF_TM "$PRIMER_PAIR_MAX_DIFF_TM" config.txt
297 #
298 # Run pal_finder
299 perl $PALFINDER_SCRIPT_DIR/pal_finder_v0.02.04.pl config.txt 2>&1 | tee pal_finder.log
300 #
301 # Check that log ends with "Done!!" message
302 if [ -z "$(tail -n 1 pal_finder.log | grep Done!!)" ] ; then
303 echo ERROR pal_finder failed to complete successfully >&2
304 exit 1
305 fi
306 #
307 # Run the pal_finder_filter.pl script from Graeme Fox
308 if [ ! -z "$FILTERED_MICROSATS" ] ; then
309 perl $PALFINDER_FILTER_PL Output/PAL_summary.txt 2>&1
310 if [ $? -ne 0 ] ; then
311 echo ERROR pal_finder_filter.pl exited with non-zero status >&2
312 exit 1
313 elif [ ! -f pal_finder_filter_output.txt ] ; then
314 echo ERROR no output from pal_finder_filter.pl >&2
315 exit 1
316 fi
317 fi
318 #
319 # Clean up
320 if [ -f Output/microsat_summary.txt ] ; then
321 /bin/mv Output/microsat_summary.txt $MICROSAT_SUMMARY
322 fi
323 if [ -f Output/PAL_summary.txt ] ; then
324 /bin/mv Output/PAL_summary.txt $PAL_SUMMARY
325 fi
326 if [ ! -z "$FILTERED_MICROSATS" ] && [ -f pal_finder_filter_output.txt ] ; then
327 echo Moving pal_finder_filter_output.txt to $FILTERED_MICROSATS
328 /bin/mv pal_finder_filter_output.txt $FILTERED_MICROSATS
329 fi
330 if [ ! -z "$OUTPUT_CONFIG_FILE" ] && [ -f config.txt ] ; then
331 /bin/mv config.txt $OUTPUT_CONFIG_FILE
332 fi
333 ##
334 #