Mercurial > repos > pjbriggs > pal_finder
annotate pal_finder_wrapper.sh @ 2:b6ccc7dd7b02 draft
Version 0.02.04.3.
| author | pjbriggs |
|---|---|
| date | Fri, 04 Dec 2015 07:43:30 -0500 |
| parents | 771ebe02636f |
| children | e1a14ed7a9d6 |
| rev | line source |
|---|---|
| 0 | 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 | |
| 2 | 29 # --filter_microsats FNAME: write output of filter options FNAME |
| 30 # -assembly FNAME: run the 'assembly' filter option and write to FNAME | |
| 31 # -primers: run the 'primers' filter option | |
| 32 # -occurrences: run the 'occurrences' filter option | |
| 33 # -rankmotifs: run the 'rankmotifs' filter option | |
| 0 | 34 # |
| 35 # pal_finder is available from http://sourceforge.net/projects/palfinder/ | |
| 36 # | |
| 37 # primer3 is available from http://primer3.sourceforge.net/releases.php | |
| 38 # (nb needs version 2.0.0-alpha) | |
| 39 # | |
| 40 # Explicitly set the locations of the pal_finder script, data files and the primer3 | |
| 41 # executable by setting the following variables in the environment: | |
| 42 # | |
| 43 # * PALFINDER_SCRIPT_DIR: location of the pal_finder Perl script (defaults to | |
| 44 # /usr/bin) | |
| 45 # * PALFINDER_DATA_DIR: location of the pal_finder data files (specifically | |
| 46 # config.txt and simple.ref; defaults to /usr/share/pal_finder_v0.02.04) | |
| 47 # * PRIMER3_CORE_EXE: name of the primer3_core program, which should include the | |
| 48 # full path if it's not on the Galaxy user's PATH (defaults to primer3_core) | |
| 49 # | |
|
1
771ebe02636f
Uploaded version 0.02.04.2: fix bug that causes tool to fail when prefix includes spaces; add explicit dependency on Perl 5.16.3.
pjbriggs
parents:
0
diff
changeset
|
50 echo "### $(basename $0) ###" |
| 0 | 51 echo $* |
| 52 # | |
| 53 # Initialise locations of scripts, data and executables | |
| 54 # | |
| 55 # Set these in the environment to overide at execution time | |
| 56 : ${PALFINDER_SCRIPT_DIR:=/usr/bin} | |
| 57 : ${PALFINDER_DATA_DIR:=/usr/share/pal_finder_v0.02.04} | |
| 58 : ${PRIMER3_CORE_EXE:=primer3_core} | |
| 59 # | |
| 60 # Filter script is in the same directory as this script | |
| 2 | 61 PALFINDER_FILTER=$(dirname $0)/pal_finder_filter_and_assembly.py |
| 62 if [ ! -f $PALFINDER_FILTER ] ; then | |
| 63 echo No $PALFINDER_FILTER script >&2 | |
| 0 | 64 exit 1 |
| 65 fi | |
| 66 # | |
| 67 # Check that we have all the components | |
| 68 function have_program() { | |
| 69 local program=$1 | |
| 70 local got_program=$(which $program 2>&1 | grep "no $(basename $program) in") | |
| 71 if [ -z "$got_program" ] ; then | |
| 72 echo yes | |
| 73 else | |
| 74 echo no | |
| 75 fi | |
| 76 } | |
| 77 if [ "$(have_program $PRIMER3_CORE_EXE)" == "no" ] ; then | |
| 78 echo "ERROR primer3_core missing: ${PRIMER3_CORE_EXE} not found" >&2 | |
| 79 exit 1 | |
| 80 fi | |
| 81 if [ ! -f "${PALFINDER_DATA_DIR}/config.txt" ] ; then | |
| 82 echo "ERROR pal_finder config.txt not found in ${PALFINDER_DATA_DIR}" >&2 | |
| 83 exit 1 | |
| 84 fi | |
| 85 if [ ! -f "${PALFINDER_SCRIPT_DIR}/pal_finder_v0.02.04.pl" ] ; then | |
| 86 echo "ERROR pal_finder_v0.02.04.pl not found in ${PALFINDER_SCRIPT_DIR}" >&2 | |
| 87 exit 1 | |
| 88 fi | |
| 89 # | |
| 90 # Initialise parameters used in the config.txt file | |
| 91 PRIMER_PREFIX="test" | |
| 92 MIN_2_MER_REPS=6 | |
| 93 MIN_3_MER_REPS=0 | |
| 94 MIN_4_MER_REPS=0 | |
| 95 MIN_5_MER_REPS=0 | |
| 96 MIN_6_MER_REPS=0 | |
| 97 PRIMER_MISPRIMING_LIBRARY=$PALFINDER_DATA_DIR/simple.ref | |
| 98 PRIMER_OPT_SIZE= | |
| 99 PRIMER_MAX_SIZE= | |
| 100 PRIMER_MIN_SIZE= | |
| 101 PRIMER_MAX_GC= | |
| 102 PRIMER_MIN_GC= | |
| 103 PRIMER_GC_CLAMP= | |
| 104 PRIMER_MAX_END_GC= | |
| 105 PRIMER_OPT_TM= | |
| 106 PRIMER_MAX_TM= | |
| 107 PRIMER_MIN_TM= | |
| 108 PRIMER_PAIR_MAX_DIFF_TM= | |
| 109 OUTPUT_CONFIG_FILE= | |
| 2 | 110 OUTPUT_ASSEMBLY= |
| 0 | 111 FILTERED_MICROSATS= |
| 2 | 112 FILTER_OPTIONS= |
| 0 | 113 # |
| 114 # Collect command line arguments | |
| 115 if [ $# -lt 2 ] ; then | |
| 116 echo "Usage: $0 FASTQ_R1 FASTQ_R2 MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]" | |
| 117 echo " $0 --454 FASTA MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]" | |
| 118 exit | |
| 119 fi | |
| 120 if [ "$1" == "--454" ] ; then | |
| 121 PLATFORM="454" | |
| 122 FNA=$2 | |
| 123 else | |
| 124 PLATFORM="Illumina" | |
| 125 FASTQ_R1=$1 | |
| 126 FASTQ_R2=$2 | |
| 127 fi | |
| 128 MICROSAT_SUMMARY=$3 | |
| 129 PAL_SUMMARY=$4 | |
| 130 shift; shift; shift; shift | |
| 131 # | |
| 132 # Collect command line options | |
| 133 while [ ! -z "$1" ] ; do | |
| 134 case "$1" in | |
| 135 --primer-prefix) | |
| 136 shift | |
| 2 | 137 # Convert all non-alphanumeric characters to underscores in prefix |
| 138 PRIMER_PREFIX=$(echo -n $1 | tr -s -c "[:alnum:]" "_") | |
| 0 | 139 ;; |
| 140 --2merMinReps) | |
| 141 shift | |
| 142 MIN_2_MER_REPS=$1 | |
| 143 ;; | |
| 144 --3merMinReps) | |
| 145 shift | |
| 146 MIN_3_MER_REPS=$1 | |
| 147 ;; | |
| 148 --4merMinReps) | |
| 149 shift | |
| 150 MIN_4_MER_REPS=$1 | |
| 151 ;; | |
| 152 --5merMinReps) | |
| 153 shift | |
| 154 MIN_5_MER_REPS=$1 | |
| 155 ;; | |
| 156 --6merMinReps) | |
| 157 shift | |
| 158 MIN_6_MER_REPS=$1 | |
| 159 ;; | |
| 160 --primer-mispriming-library) | |
| 161 shift | |
| 162 PRIMER_MISPRIMING_LIBRARY=$1 | |
| 163 ;; | |
| 164 --primer-opt-size) | |
| 165 shift | |
| 166 PRIMER_OPT_SIZE=$1 | |
| 167 ;; | |
| 168 --primer-max-size) | |
| 169 shift | |
| 170 PRIMER_MAX_SIZE=$1 | |
| 171 ;; | |
| 172 --primer-min-size) | |
| 173 shift | |
| 174 PRIMER_MIN_SIZE=$1 | |
| 175 ;; | |
| 176 --primer-max-gc) | |
| 177 shift | |
| 178 PRIMER_MAX_GC=$1 | |
| 179 ;; | |
| 180 --primer-min-gc) | |
| 181 shift | |
| 182 PRIMER_MIN_GC=$1 | |
| 183 ;; | |
| 184 --primer-gc-clamp) | |
| 185 shift | |
| 186 PRIMER_GC_CLAMP=$1 | |
| 187 ;; | |
| 188 --primer-max-end-gc) | |
| 189 shift | |
| 190 PRIMER_MAX_END_GC=$1 | |
| 191 ;; | |
| 192 --primer-opt-tm) | |
| 193 shift | |
| 194 PRIMER_OPT_TM=$1 | |
| 195 ;; | |
| 196 --primer-max-tm) | |
| 197 shift | |
| 198 PRIMER_MAX_TM=$1 | |
| 199 ;; | |
| 200 --primer-min-tm) | |
| 201 shift | |
| 202 PRIMER_MIN_TM=$1 | |
| 203 ;; | |
| 204 --primer-pair-max-diff-tm) | |
| 205 shift | |
| 206 PRIMER_PAIR_MAX_DIFF_TM=$1 | |
| 207 ;; | |
| 208 --output_config_file) | |
| 209 shift | |
| 210 OUTPUT_CONFIG_FILE=$1 | |
| 211 ;; | |
| 212 --filter_microsats) | |
| 213 shift | |
| 214 FILTERED_MICROSATS=$1 | |
| 215 ;; | |
| 2 | 216 -primers|-occurrences|-rankmotifs) |
| 217 FILTER_OPTIONS="$FILTER_OPTIONS $1" | |
| 218 ;; | |
| 219 -assembly) | |
| 220 FILTER_OPTIONS="$FILTER_OPTIONS $1" | |
| 221 shift | |
| 222 OUTPUT_ASSEMBLY=$1 | |
| 223 ;; | |
| 0 | 224 *) |
| 225 echo Unknown option: $1 >&2 | |
| 226 exit 1 | |
| 227 ;; | |
| 228 esac | |
| 229 shift | |
| 230 done | |
| 231 # | |
| 232 # Check that primer3_core is available | |
| 233 got_primer3=`which $PRIMER3_CORE_EXE 2>&1 | grep -v "no primer3_core in"` | |
| 234 if [ -z "$got_primer3" ] ; then | |
| 235 echo ERROR primer3_core not found >&2 | |
| 236 exit 1 | |
| 237 fi | |
| 238 # | |
| 239 # Set up the working dir | |
| 240 if [ "$PLATFORM" == "Illumina" ] ; then | |
| 241 # Paired end Illumina data as input | |
| 242 if [ $FASTQ_R1 == $FASTQ_R2 ] ; then | |
| 243 echo ERROR R1 and R2 fastqs are the same file >&2 | |
| 244 exit 1 | |
| 245 fi | |
| 246 ln -s $FASTQ_R1 | |
| 247 ln -s $FASTQ_R2 | |
| 248 fastq_r1=$(basename $FASTQ_R1) | |
| 249 fastq_r2=$(basename $FASTQ_R2) | |
| 250 else | |
| 251 # 454 data as input | |
| 252 ln -s $FNA | |
| 253 fna=$(basename $FNA) | |
| 254 fi | |
| 255 ln -s $PRIMER_MISPRIMING_LIBRARY | |
| 256 PRIMER_MISPRIMING_LIBRARY=$(basename $PRIMER_MISPRIMING_LIBRARY) | |
| 257 mkdir Output | |
| 258 # | |
| 259 # Copy in the default config.txt file | |
|
1
771ebe02636f
Uploaded version 0.02.04.2: fix bug that causes tool to fail when prefix includes spaces; add explicit dependency on Perl 5.16.3.
pjbriggs
parents:
0
diff
changeset
|
260 echo "### Creating config.txt file for pal_finder run ###" |
| 0 | 261 /bin/cp $PALFINDER_DATA_DIR/config.txt . |
| 262 # | |
| 263 # Update the config.txt file with new values | |
| 264 function set_config_value() { | |
| 265 local key=$1 | |
| 266 local value=$2 | |
| 267 local config_txt=$3 | |
| 268 if [ -z "$value" ] ; then | |
| 269 echo "No value for $key, left as default" | |
| 270 else | |
| 271 echo Setting "$key" to "$value" | |
| 272 sed -i 's,^'"$key"' .*,'"$key"' '"$value"',' $config_txt | |
| 273 fi | |
| 274 } | |
| 275 # Input files | |
| 276 set_config_value platform $PLATFORM config.txt | |
| 277 if [ "$PLATFORM" == "Illumina" ] ; then | |
| 278 set_config_value inputFormat fastq config.txt | |
| 279 set_config_value pairedEnd 1 config.txt | |
| 280 set_config_value inputReadFile $fastq_r1 config.txt | |
| 281 set_config_value pairedReadFile $fastq_r2 config.txt | |
| 282 else | |
| 283 set_config_value inputFormat fasta config.txt | |
| 284 set_config_value pairedEnd 0 config.txt | |
| 285 set_config_value input454reads $fna config.txt | |
| 286 fi | |
| 287 # Output files | |
| 288 set_config_value MicrosatSumOut Output/microsat_summary.txt config.txt | |
| 289 set_config_value PALsummaryOut Output/PAL_summary.txt config.txt | |
| 290 # Microsat info | |
| 291 set_config_value 2merMinReps $MIN_2_MER_REPS config.txt | |
| 292 set_config_value 3merMinReps $MIN_3_MER_REPS config.txt | |
| 293 set_config_value 4merMinReps $MIN_4_MER_REPS config.txt | |
| 294 set_config_value 5merMinReps $MIN_5_MER_REPS config.txt | |
| 295 set_config_value 6merMinReps $MIN_6_MER_REPS config.txt | |
| 296 # Primer3 settings | |
| 297 set_config_value primer3input Output/pr3in.txt config.txt | |
| 298 set_config_value primer3output Output/pr3out.txt config.txt | |
| 299 set_config_value primer3executable $PRIMER3_CORE_EXE config.txt | |
| 300 set_config_value prNamePrefix ${PRIMER_PREFIX}_ config.txt | |
| 301 set_config_value PRIMER_MISPRIMING_LIBRARY "$PRIMER_MISPRIMING_LIBRARY" config.txt | |
| 302 set_config_value PRIMER_OPT_SIZE "$PRIMER_OPT_SIZE" config.txt | |
| 303 set_config_value PRIMER_MIN_SIZE "$PRIMER_MIN_SIZE" config.txt | |
| 304 set_config_value PRIMER_MAX_SIZE "$PRIMER_MAX_SIZE" config.txt | |
| 305 set_config_value PRIMER_MIN_GC "$PRIMER_MIN_GC" config.txt | |
| 306 set_config_value PRIMER_MAX_GC "$PRIMER_MAX_GC" config.txt | |
| 307 set_config_value PRIMER_GC_CLAMP "$PRIMER_GC_CLAMP" config.txt | |
| 308 set_config_value PRIMER_MAX_END_GC "$PRIMER_MAX_END_GC" config.txt | |
| 309 set_config_value PRIMER_MIN_TM "$PRIMER_MIN_TM" config.txt | |
| 310 set_config_value PRIMER_MAX_TM "$PRIMER_MAX_TM" config.txt | |
| 311 set_config_value PRIMER_OPT_TM "$PRIMER_OPT_TM" config.txt | |
| 312 set_config_value PRIMER_PAIR_MAX_DIFF_TM "$PRIMER_PAIR_MAX_DIFF_TM" config.txt | |
| 313 # | |
| 314 # Run pal_finder | |
|
1
771ebe02636f
Uploaded version 0.02.04.2: fix bug that causes tool to fail when prefix includes spaces; add explicit dependency on Perl 5.16.3.
pjbriggs
parents:
0
diff
changeset
|
315 echo "### Running pal_finder ###" |
| 0 | 316 perl $PALFINDER_SCRIPT_DIR/pal_finder_v0.02.04.pl config.txt 2>&1 | tee pal_finder.log |
|
1
771ebe02636f
Uploaded version 0.02.04.2: fix bug that causes tool to fail when prefix includes spaces; add explicit dependency on Perl 5.16.3.
pjbriggs
parents:
0
diff
changeset
|
317 echo "### pal_finder finised ###" |
| 0 | 318 # |
| 319 # Check that log ends with "Done!!" message | |
| 320 if [ -z "$(tail -n 1 pal_finder.log | grep Done!!)" ] ; then | |
| 321 echo ERROR pal_finder failed to complete successfully >&2 | |
| 322 exit 1 | |
| 323 fi | |
| 324 # | |
| 2 | 325 # Sort outputs into a consistent order regardless of Perl version |
| 326 echo "### Sorting outputs ###" | |
| 327 head -1 Output/PAL_summary.txt > Output/PAL_summary.sorted.txt | |
| 328 if [ "$PLATFORM" == "Illumina" ] ; then | |
| 329 grep -v "^readPairID" Output/PAL_summary.txt | sort -k 1 >> Output/PAL_summary.sorted.txt | |
| 330 else | |
| 331 grep -v "^SequenceID" Output/PAL_summary.txt | sort -k 1 >> Output/PAL_summary.sorted.txt | |
| 332 fi | |
| 333 mv Output/PAL_summary.sorted.txt Output/PAL_summary.txt | |
| 334 # | |
| 335 # Run the filtering & assembly script | |
| 336 if [ ! -z "$FILTERED_MICROSATS" ] || [ ! -z "$OUTPUT_ASSEMBLY" ] ; then | |
| 337 echo "### Running filtering & assembly script ###" | |
| 338 python $PALFINDER_FILTER -i $fastq_r1 -j $fastq_r2 -p Output/PAL_summary.txt $FILTER_OPTIONS 2>&1 | |
| 0 | 339 if [ $? -ne 0 ] ; then |
| 2 | 340 echo ERROR $PALFINDER_FILTER exited with non-zero status >&2 |
| 0 | 341 exit 1 |
| 2 | 342 elif [ ! -f PAL_summary.filtered ] ; then |
| 343 echo ERROR no output from $PALFINDER_FILTER >&2 | |
| 0 | 344 exit 1 |
| 345 fi | |
| 346 fi | |
| 347 # | |
| 348 # Clean up | |
|
1
771ebe02636f
Uploaded version 0.02.04.2: fix bug that causes tool to fail when prefix includes spaces; add explicit dependency on Perl 5.16.3.
pjbriggs
parents:
0
diff
changeset
|
349 echo "### Handling output files ###" |
| 0 | 350 if [ -f Output/microsat_summary.txt ] ; then |
| 351 /bin/mv Output/microsat_summary.txt $MICROSAT_SUMMARY | |
| 352 fi | |
| 353 if [ -f Output/PAL_summary.txt ] ; then | |
| 354 /bin/mv Output/PAL_summary.txt $PAL_SUMMARY | |
| 355 fi | |
| 2 | 356 if [ ! -z "$FILTERED_MICROSATS" ] && [ -f PAL_summary.filtered ] ; then |
| 357 /bin/mv PAL_summary.filtered $FILTERED_MICROSATS | |
| 358 fi | |
| 359 if [ ! -z "$OUTPUT_ASSEMBLY" ] ; then | |
| 360 assembly=${fastq_r1%.*}_pal_finder_assembly_output.txt | |
| 361 if [ -f "$assembly" ] ; then | |
| 362 /bin/mv $assembly "$OUTPUT_ASSEMBLY" | |
| 363 fi | |
| 0 | 364 fi |
| 365 if [ ! -z "$OUTPUT_CONFIG_FILE" ] && [ -f config.txt ] ; then | |
| 366 /bin/mv config.txt $OUTPUT_CONFIG_FILE | |
| 367 fi | |
| 368 ## | |
| 369 # |