Mercurial > repos > pjbriggs > pal_finder

comparison pal_finder_wrapper.sh @ 9:b3ca3ece8a8b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
WIP detect/trap "bad primer product range" errors.
author pjbriggs
date Fri, 16 Jun 2017 09:12:09 -0400
parents 67ab365c29a7
children 84613d5f1713
comparison
equal deleted inserted replaced
8:67ab365c29a7 9:b3ca3ece8a8b
51 echo $* 51 echo $*
52 # 52 #
53 # Maximum size reporting log file contents 53 # Maximum size reporting log file contents
54 MAX_LINES=500 54 MAX_LINES=500
55 # 55 #
56 # Get helper functions
57 . $(dirname $0)/pal_finder_wrapper_utils.sh
58 #
56 # Initialise locations of scripts, data and executables 59 # Initialise locations of scripts, data and executables
57 # 60 #
58 # Set these in the environment to overide at execution time 61 # Set these in the environment to overide at execution time
59 : ${PALFINDER_SCRIPT_DIR:=/usr/bin} 62 : ${PALFINDER_SCRIPT_DIR:=/usr/bin}
60 : ${PALFINDER_DATA_DIR:=/usr/share/pal_finder_v0.02.04} 63 : ${PALFINDER_DATA_DIR:=/usr/share/pal_finder_v0.02.04}
61 : ${PRIMER3_CORE_EXE:=primer3_core} 64 : ${PRIMER3_CORE_EXE:=primer3_core}
62 # 65 #
63 # Filter script is in the same directory as this script 66 # Filter script is in the same directory as this script
64 PALFINDER_FILTER=$(dirname $0)/pal_filter.py 67 PALFINDER_FILTER=$(dirname $0)/pal_filter.py
65 if [ ! -f $PALFINDER_FILTER ] ; then 68 if [ ! -f $PALFINDER_FILTER ] ; then
66 echo No $PALFINDER_FILTER script >&2 69 fatal No $PALFINDER_FILTER script
67 exit 1
68 fi 70 fi
69 # 71 #
70 # Check that we have all the components 72 # Check that we have all the components
71 function have_program() {
72 local program=$1
73 local got_program=$(which $program 2>&1 | grep "no $(basename $program) in")
74 if [ -z "$got_program" ] ; then
75 echo yes
76 else
77 echo no
78 fi
79 }
80 if [ "$(have_program $PRIMER3_CORE_EXE)" == "no" ] ; then 73 if [ "$(have_program $PRIMER3_CORE_EXE)" == "no" ] ; then
81 echo "ERROR primer3_core missing: ${PRIMER3_CORE_EXE} not found" >&2 74 fatal "primer3_core missing: ${PRIMER3_CORE_EXE} not found"
82 exit 1
83 fi 75 fi
84 if [ ! -f "${PALFINDER_DATA_DIR}/config.txt" ] ; then 76 if [ ! -f "${PALFINDER_DATA_DIR}/config.txt" ] ; then
85 echo "ERROR pal_finder config.txt not found in ${PALFINDER_DATA_DIR}" >&2 77 fatal "pal_finder config.txt not found in ${PALFINDER_DATA_DIR}"
86 exit 1
87 fi 78 fi
88 if [ ! -f "${PALFINDER_SCRIPT_DIR}/pal_finder_v0.02.04.pl" ] ; then 79 if [ ! -f "${PALFINDER_SCRIPT_DIR}/pal_finder_v0.02.04.pl" ] ; then
89 echo "ERROR pal_finder_v0.02.04.pl not found in ${PALFINDER_SCRIPT_DIR}" >&2 80 fatal "pal_finder_v0.02.04.pl not found in ${PALFINDER_SCRIPT_DIR}"
90 exit 1
91 fi 81 fi
92 # 82 #
93 # Initialise parameters used in the config.txt file 83 # Initialise parameters used in the config.txt file
94 PRIMER_PREFIX="test" 84 PRIMER_PREFIX="test"
95 MIN_2_MER_REPS=6 85 MIN_2_MER_REPS=6
116 # 106 #
117 # Collect command line arguments 107 # Collect command line arguments
118 if [ $# -lt 2 ] ; then 108 if [ $# -lt 2 ] ; then
119 echo "Usage: $0 FASTQ_R1 FASTQ_R2 MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]" 109 echo "Usage: $0 FASTQ_R1 FASTQ_R2 MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]"
120 echo " $0 --454 FASTA MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]" 110 echo " $0 --454 FASTA MICROSAT_SUMMARY PAL_SUMMARY [OPTIONS]"
121 exits 111 fatal "Bad command line"
122 fi 112 fi
123 if [ "$1" == "--454" ] ; then 113 if [ "$1" == "--454" ] ; then
124 PLATFORM="454" 114 PLATFORM="454"
125 FNA=$2 115 FNA=$2
126 else 116 else
233 done 223 done
234 # 224 #
235 # Check that primer3_core is available 225 # Check that primer3_core is available
236 got_primer3=`which $PRIMER3_CORE_EXE 2>&1 | grep -v "no primer3_core in"` 226 got_primer3=`which $PRIMER3_CORE_EXE 2>&1 | grep -v "no primer3_core in"`
237 if [ -z "$got_primer3" ] ; then 227 if [ -z "$got_primer3" ] ; then
238 echo ERROR primer3_core not found >&2 228 fatal "primer3_core not found"
239 exit 1
240 fi 229 fi
241 # 230 #
242 # Set up the working dir 231 # Set up the working dir
243 if [ "$PLATFORM" == "Illumina" ] ; then 232 if [ "$PLATFORM" == "Illumina" ] ; then
244 # Paired end Illumina data as input 233 # Paired end Illumina data as input
245 if [ $FASTQ_R1 == $FASTQ_R2 ] ; then 234 if [ $FASTQ_R1 == $FASTQ_R2 ] ; then
246 echo ERROR R1 and R2 fastqs are the same file >&2 235 fatal ERROR R1 and R2 fastqs are the same file
247 exit 1
248 fi 236 fi
249 ln -s $FASTQ_R1 237 ln -s $FASTQ_R1
250 ln -s $FASTQ_R2 238 ln -s $FASTQ_R2
251 fastq_r1=$(basename $FASTQ_R1) 239 fastq_r1=$(basename $FASTQ_R1)
252 fastq_r2=$(basename $FASTQ_R2) 240 fastq_r2=$(basename $FASTQ_R2)
262 # Copy in the default config.txt file 250 # Copy in the default config.txt file
263 echo "### Creating config.txt file for pal_finder run ###" 251 echo "### Creating config.txt file for pal_finder run ###"
264 /bin/cp $PALFINDER_DATA_DIR/config.txt . 252 /bin/cp $PALFINDER_DATA_DIR/config.txt .
265 # 253 #
266 # Update the config.txt file with new values 254 # Update the config.txt file with new values
267 function set_config_value() {
268 local key=$1
269 local value=$2
270 local config_txt=$3
271 if [ -z "$value" ] ; then
272 echo "No value for $key, left as default"
273 else
274 echo Setting "$key" to "$value"
275 sed -i 's,^'"$key"' .*,'"$key"' '"$value"',' $config_txt
276 fi
277 }
278 # Input files 255 # Input files
279 set_config_value platform $PLATFORM config.txt 256 set_config_value platform $PLATFORM config.txt
280 if [ "$PLATFORM" == "Illumina" ] ; then 257 if [ "$PLATFORM" == "Illumina" ] ; then
281 set_config_value inputFormat fastq config.txt 258 set_config_value inputFormat fastq config.txt
282 set_config_value pairedEnd 1 config.txt 259 set_config_value pairedEnd 1 config.txt
297 set_config_value 5merMinReps $MIN_5_MER_REPS config.txt 274 set_config_value 5merMinReps $MIN_5_MER_REPS config.txt
298 set_config_value 6merMinReps $MIN_6_MER_REPS config.txt 275 set_config_value 6merMinReps $MIN_6_MER_REPS config.txt
299 # Primer3 settings 276 # Primer3 settings
300 set_config_value primer3input Output/pr3in.txt config.txt 277 set_config_value primer3input Output/pr3in.txt config.txt
301 set_config_value primer3output Output/pr3out.txt config.txt 278 set_config_value primer3output Output/pr3out.txt config.txt
279 set_config_value keepPrimer3files 1 config.txt
302 set_config_value primer3executable $PRIMER3_CORE_EXE config.txt 280 set_config_value primer3executable $PRIMER3_CORE_EXE config.txt
303 set_config_value prNamePrefix ${PRIMER_PREFIX}_ config.txt 281 set_config_value prNamePrefix ${PRIMER_PREFIX}_ config.txt
304 set_config_value PRIMER_MISPRIMING_LIBRARY "$PRIMER_MISPRIMING_LIBRARY" config.txt 282 set_config_value PRIMER_MISPRIMING_LIBRARY "$PRIMER_MISPRIMING_LIBRARY" config.txt
305 set_config_value PRIMER_OPT_SIZE "$PRIMER_OPT_SIZE" config.txt 283 set_config_value PRIMER_OPT_SIZE "$PRIMER_OPT_SIZE" config.txt
306 set_config_value PRIMER_MIN_SIZE "$PRIMER_MIN_SIZE" config.txt 284 set_config_value PRIMER_MIN_SIZE "$PRIMER_MIN_SIZE" config.txt
327 fi 305 fi
328 tail -$MAX_LINES pal_finder.log 306 tail -$MAX_LINES pal_finder.log
329 # 307 #
330 # Check that log ends with "Done!!" message 308 # Check that log ends with "Done!!" message
331 if [ -z "$(tail -n 1 pal_finder.log | grep Done!!)" ] ; then 309 if [ -z "$(tail -n 1 pal_finder.log | grep Done!!)" ] ; then
332 echo ERROR pal_finder failed to complete successfully >&2 310 fatal ERROR pal_finder failed to complete successfully
333 exit 1 311 fi
312 echo "### pal_finder finished ###"
313 #
314 # Check for errors in pal_finder output
315 echo "### Checking for errors ###"
316 if [ ! -z "$(grep 'primer3_core: Illegal element in PRIMER_PRODUCT_SIZE_RANGE' pal_finder.log)" ] ; then
317 echo ERROR primer3 terminated prematurely due to bad product size ranges
318 cat >&2 <<EOF
319 ERROR primer3 terminated prematurely due to bad product size ranges
320
321 Pal_finder generated bad ranges for the following read IDs:
322 EOF
323 $(find_bad_primer_ranges Output/pr3in.txt) >&2
324 cat >&2 <<EOF
325
326 This error can occur when input data contains short R1 reads and has
327 has not been properly trimmed and filtered.
328
329 EOF
330 fatal pal_finder failed to complete successfully
331 EOF
334 fi 332 fi
335 # 333 #
336 # Sort outputs into a consistent order regardless of Perl version 334 # Sort outputs into a consistent order regardless of Perl version
337 echo "### Sorting outputs ###" 335 echo "### Sorting outputs ###"
338 head -1 Output/PAL_summary.txt > Output/PAL_summary.sorted.txt 336 head -1 Output/PAL_summary.txt > Output/PAL_summary.sorted.txt
352 echo WARNING output too long, truncated to last $MAX_LINES lines: 350 echo WARNING output too long, truncated to last $MAX_LINES lines:
353 echo ... 351 echo ...
354 fi 352 fi
355 tail -$MAX_LINES pal_filter.log 353 tail -$MAX_LINES pal_filter.log
356 if [ $? -ne 0 ] ; then 354 if [ $? -ne 0 ] ; then
357 echo ERROR $PALFINDER_FILTER exited with non-zero status >&2 355 fatal $PALFINDER_FILTER exited with non-zero status
358 exit 1
359 elif [ ! -f PAL_summary.filtered ] ; then 356 elif [ ! -f PAL_summary.filtered ] ; then
360 echo ERROR no output from $PALFINDER_FILTER >&2 357 fatal no output from $PALFINDER_FILTER
361 exit 1
362 fi 358 fi
363 fi 359 fi
364 # 360 #
365 # Clean up 361 # Clean up
366 echo "### Handling output files ###" 362 echo "### Handling output files ###"
376 if [ ! -z "$OUTPUT_ASSEMBLY" ] ; then 372 if [ ! -z "$OUTPUT_ASSEMBLY" ] ; then
377 assembly=${fastq_r1%.*}_pal_filter_assembly_output.txt 373 assembly=${fastq_r1%.*}_pal_filter_assembly_output.txt
378 if [ -f "$assembly" ] ; then 374 if [ -f "$assembly" ] ; then
379 /bin/mv $assembly "$OUTPUT_ASSEMBLY" 375 /bin/mv $assembly "$OUTPUT_ASSEMBLY"
380 else 376 else
381 echo ERROR no assembly output found >&2 377 fatal no assembly output found
382 exit 1
383 fi 378 fi
384 fi 379 fi
385 if [ ! -z "$OUTPUT_CONFIG_FILE" ] && [ -f config.txt ] ; then 380 if [ ! -z "$OUTPUT_CONFIG_FILE" ] && [ -f config.txt ] ; then
386 /bin/mv config.txt $OUTPUT_CONFIG_FILE 381 /bin/mv config.txt $OUTPUT_CONFIG_FILE
387 fi 382 fi