added useful biopiece scripts

git-svn-id: http://biopieces.googlecode.com/svn/trunk@1347 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_scripts/remove_adaptor.sh b/bp_scripts/remove_adaptor.sh
new file mode 100755 (executable)
index 0000000..daf036b
--- /dev/null
+++ b/bp_scripts/remove_adaptor.sh
@@ -0,0 +1,15 @@
+#!/bin/sh
+
+# Remove adaptor from sequences and produce a number of plots.
+
+# Usage: read_fastq -i test.fq | remove_adaptor.sh | write_fastq -xo test_no_adaptor.fq
+
+adaptor='AGATCGGAAGACACACGTCT'  # Solexa adaptor
+#adaptor='TCGTATGCCGTCTTCTGCTTG' # 454 adaptor
+pid=$$
+
+parallel -k --blocksize 50M --pipe --recend "\n---\n" "nice -n 19 find_adaptor -a $adaptor" |
+plot_histogram -t post -o remove_adaptor_histogram.$pid.ps -k ADAPTOR_POS -s num |
+plot_lendist -t post -o remove_adaptor_lendist.$pid.ps -k ADAPTOR_LEN |
+analyze_vals -k ADAPTOR_POS,ADAPTOR_LEN -o remove_adaptor_analyze_vals.$pid.txt |
+clip_adaptor

diff --git a/bp_scripts/trim_seq.sh b/bp_scripts/trim_seq.sh
new file mode 100755 (executable)
index 0000000..60839aa
--- /dev/null
+++ b/bp_scripts/trim_seq.sh
@@ -0,0 +1,15 @@
+#!/bin/sh
+
+# Trim sequences in the stream based on quality scores and
+# outputs a number of plots.
+
+pid=$$
+
+plot_scores -t post -o trim_seq_scores_pretrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o trim_seq_lendist_pretrim.$pid.ps |
+trim_seq |
+grab -e "SEQ_LEN>=30" |
+mean_scores -l |
+grab -e "SCORES_LOCAL_MEAN>=15" |
+plot_scores -t post -o trim_seq_scores_posttrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o trim_seq_lendist_posttrim.$pid.ps