working on clean_reads scripts

git-svn-id: http://biopieces.googlecode.com/svn/trunk@1424 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_scripts/clean_reads.sh b/bp_scripts/clean_reads.sh
deleted file mode 100755 (executable)
index 6929378..0000000
--- a/bp_scripts/clean_reads.sh
+++ /dev/null
@@ -1,27 +0,0 @@
-#!/bin/sh
-
-# Remove adaptor from sequences and trim according to scores.
-# Produces a number of plots during the process.
-
-# Usage: read_fastq -i test.fq | clean_reads.sh | write_fastq -xo clean.fq
-
-#adaptor='TCGTATGCCGTCTTCTGCTTG' # 454 adaptor
-adaptor='AGATCGGAAGACACACGTCT'  # Solexa adaptor
-pid=$$
-
-progress_meter |
-plot_lendist -t post -o 00_remove_adaptor_lendist_seq_before.$pid.ps -k SEQ_LEN |
-find_adaptor -a $adaptor -p |
-plot_histogram -t post -o 01_remove_adaptor_histogram_adaptor_pos.$pid.ps -k ADAPTOR_POS -s num |
-plot_lendist -t post -o 02_remove_adaptor_lendist_adaptor_len.$pid.ps -k ADAPTOR_LEN |
-plot_lendist -t post -o 03_remove_adaptor_lendist_seq_after.$pid.ps -k SEQ_LEN |
-analyze_vals -k ADAPTOR_POS,ADAPTOR_LEN -o 04_remove_adaptor_analyze_vals.$pid.txt |
-clip_adaptor |
-plot_scores -t post -o 04_trim_seq_scores_pretrim.$pid.ps |
-plot_lendist -k SEQ_LEN -t post -o 05_trim_seq_lendist_pretrim.$pid.ps |
-trim_seq |
-grab -e "SEQ_LEN>=30" |
-mean_scores -l |
-grab -e "SCORES_LOCAL_MEAN>=15" |
-plot_scores -t post -o 06_trim_seq_scores_posttrim.$pid.ps |
-plot_lendist -k SEQ_LEN -t post -o 07_trim_seq_lendist_posttrim.$pid.ps

diff --git a/bp_scripts/clean_reads_454.sh b/bp_scripts/clean_reads_454.sh
new file mode 100755 (executable)
index 0000000..70e05e9
--- /dev/null
+++ b/bp_scripts/clean_reads_454.sh
@@ -0,0 +1,27 @@
+#!/bin/sh
+
+# Remove adaptor from sequences and trim according to scores.
+# Produces a number of plots during the process.
+
+# Usage: read_fastq -i test.fq | clean_reads.sh | write_fastq -xo clean.fq
+
+adaptor='TCGTATGCCGTCTTCTGCTTG' # 454 adaptor
+#adaptor='AGATCGGAAGACACACGTCT'  # Solexa adaptor
+pid=$$
+
+progress_meter |
+plot_lendist -t post -o 00_remove_adaptor_lendist_seq_before.$pid.ps -k SEQ_LEN |
+find_adaptor -a $adaptor -p |
+plot_histogram -t post -o 01_remove_adaptor_histogram_adaptor_pos.$pid.ps -k ADAPTOR_POS -s num |
+plot_lendist -t post -o 02_remove_adaptor_lendist_adaptor_len.$pid.ps -k ADAPTOR_LEN |
+plot_lendist -t post -o 03_remove_adaptor_lendist_seq_after.$pid.ps -k SEQ_LEN |
+analyze_vals -k ADAPTOR_POS,ADAPTOR_LEN -o 04_remove_adaptor_analyze_vals.$pid.txt |
+clip_adaptor |
+plot_scores -t post -o 04_trim_seq_scores_pretrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o 05_trim_seq_lendist_pretrim.$pid.ps |
+trim_seq |
+grab -e "SEQ_LEN>=50" |
+mean_scores -l |
+grab -e "SCORES_LOCAL_MEAN>=15" |
+plot_scores -t post -o 06_trim_seq_scores_posttrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o 07_trim_seq_lendist_posttrim.$pid.ps

diff --git a/bp_scripts/clean_reads_illumina.sh b/bp_scripts/clean_reads_illumina.sh
new file mode 100755 (executable)
index 0000000..6929378
--- /dev/null
+++ b/bp_scripts/clean_reads_illumina.sh
@@ -0,0 +1,27 @@
+#!/bin/sh
+
+# Remove adaptor from sequences and trim according to scores.
+# Produces a number of plots during the process.
+
+# Usage: read_fastq -i test.fq | clean_reads.sh | write_fastq -xo clean.fq
+
+#adaptor='TCGTATGCCGTCTTCTGCTTG' # 454 adaptor
+adaptor='AGATCGGAAGACACACGTCT'  # Solexa adaptor
+pid=$$
+
+progress_meter |
+plot_lendist -t post -o 00_remove_adaptor_lendist_seq_before.$pid.ps -k SEQ_LEN |
+find_adaptor -a $adaptor -p |
+plot_histogram -t post -o 01_remove_adaptor_histogram_adaptor_pos.$pid.ps -k ADAPTOR_POS -s num |
+plot_lendist -t post -o 02_remove_adaptor_lendist_adaptor_len.$pid.ps -k ADAPTOR_LEN |
+plot_lendist -t post -o 03_remove_adaptor_lendist_seq_after.$pid.ps -k SEQ_LEN |
+analyze_vals -k ADAPTOR_POS,ADAPTOR_LEN -o 04_remove_adaptor_analyze_vals.$pid.txt |
+clip_adaptor |
+plot_scores -t post -o 04_trim_seq_scores_pretrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o 05_trim_seq_lendist_pretrim.$pid.ps |
+trim_seq |
+grab -e "SEQ_LEN>=30" |
+mean_scores -l |
+grab -e "SCORES_LOCAL_MEAN>=15" |
+plot_scores -t post -o 06_trim_seq_scores_posttrim.$pid.ps |
+plot_lendist -k SEQ_LEN -t post -o 07_trim_seq_lendist_posttrim.$pid.ps