Mercurial > repos > pjbriggs > pal_finder

comparison pal_finder_wrapper_utils.sh @ 8:4e625d3672ba draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Pal_finder tool version 0.02.04.7: add detection/reporting of bad ranges; enable subset of reads to be used; check n-mers.
author pjbriggs
date Wed, 16 May 2018 07:39:16 -0400
parents
children
comparison
equal deleted inserted replaced
7:5e133b7b79a6 8:4e625d3672ba
1 #!/bin/bash
2 #
3 # Helper functions for the pal_finder_wrapper.sh script
4 #
5 # Utility function for terminating on fatal error
6 function fatal() {
7 echo "FATAL $@" >&2
8 exit 1
9 }
10 #
11 # Check that specified program is available
12 function have_program() {
13 local program=$1
14 local got_program=$(which $program 2>&1 | grep "no $(basename $program) in")
15 if [ -z "$got_program" ] ; then
16 echo yes
17 else
18 echo no
19 fi
20 }
21 #
22 # Set the value for a parameter in the pal_finder config file
23 function set_config_value() {
24 local key=$1
25 local value=$2
26 local config_txt=$3
27 if [ -z "$value" ] ; then
28 echo "No value for $key, left as default"
29 else
30 echo Setting "$key" to "$value"
31 sed -i 's,^'"$key"' .*,'"$key"' '"$value"',' $config_txt
32 fi
33 }
34 #
35 # Identify 'bad' PRIMER_PRODUCT_SIZE_RANGE from pr3in.txt file
36 function find_bad_primer_ranges() {
37 # Parses a pr3in.txt file from pal_finder and reports
38 # sequence ids where the PRIMER_PRODUCT_SIZE_RANGE has
39 # upper limit which is smaller than lower limit
40 local pr3in=$1
41 local outfile=$2
42 local pattern="^(SEQUENCE_ID|PRIMER_PRODUCT_SIZE_RANGE)"
43 for line in $(grep -E "$pattern" $pr3in | sed 's/ /^/' | sed 'N;s/\n/*/')
44 do
45 # Loop over pairs of SEQUENCE_ID and PRIMER_PRODUCT_SIZE_RANGE
46 # keywords in the primer3 input
47 if [ ! -z "$(echo $line | grep ^SEQUENCE_ID)" ] ; then
48 # Lines look like:
49 # SEQUENCE_ID=(AT_1_16)(AT_1_16)M00879:99:000000000-AH9KG:1:2107:10006:2535*PRIMER_PRODUCT_SIZE_RANGE=194-329^59-194
50 local size_range=$(echo $line | cut -d'*' -f2 | cut -d'=' -f2 | tr '^' ' ')
51 local seq_id=$(echo $line | cut -d'*' -f1 | cut -d'=' -f2)
52 elif [ ! -z "$(echo $line | grep ^PRIMER_PRODUCT_SIZE_RANGE)" ] ; then
53 # Lines look like:
54 # PRIMER_PRODUCT_SIZE_RANGE=194-329^59-194*SEQUENCE_ID=(AT_1_16)(AT_1_16)M00879:99:000000000-AH9KG:1:2107:10006:2535
55 local size_range=$(echo $line | cut -d'*' -f1 | cut -d'=' -f2 | tr '^' ' ')
56 local seq_id=$(echo $line | cut -d'*' -f2 | cut -d'=' -f2)
57 fi
58 seq_id=$(echo $seq_id | cut -d')' -f3)
59 # Check the upper and lower limits in each range
60 # to see if it's okay
61 local bad_range=
62 for range in $(echo $size_range) ; do
63 local lower=$(echo $range | cut -d'-' -f1)
64 local upper=$(echo $range | cut -d'-' -f2)
65 if [ "$lower" -gt "$upper" ] ; then
66 bad_range=yes
67 break
68 fi
69 done
70 # Report if the range is wrong
71 if [ ! -z "$bad_range" ] ; then
72 echo "${seq_id}"$'\t'"(${size_range})" >>$outfile
73 fi
74 done
75 }