From 78f6ad966bc5718e517b57ab5ba2aee63c0c82f7 Mon Sep 17 00:00:00 2001
From: "Haarst, Jan van" <jan.vanhaarst@wur.nl>
Date: Thu, 3 Nov 2016 11:41:17 +0100
Subject: [PATCH] Add downloading of test data.
---
src/rules/repeats/ReAs.rules | 0
src/rules/repeats/reas.rules | 158 ++++++++++++++-----------
src/workflows/repeats/reas/Snakefile | 4 +-
src/workflows/repeats/reas/config.json | 21 +---
4 files changed, 93 insertions(+), 90 deletions(-)
delete mode 100644 src/rules/repeats/ReAs.rules
diff --git a/src/rules/repeats/ReAs.rules b/src/rules/repeats/ReAs.rules
deleted file mode 100644
index e69de29..0000000
diff --git a/src/rules/repeats/reas.rules b/src/rules/repeats/reas.rules
index e5ed539..105eb2e 100644
--- a/src/rules/repeats/reas.rules
+++ b/src/rules/repeats/reas.rules
@@ -1,72 +1,86 @@
-"""
-@version: 0.1
-
-Run ReAS repeat discovery tool
-
-Required programs:
-* ReAS
-* MUSCLE
-* BLASTN (from BLAST-suite)
-* dustmasker (from BLAST-suite)
-
-"""
-
-import os, subprocess
-import configparser
-from snakemake.remote.FTP import RemoteProvider as FTPRemoteProvider
-
-__author__ = "Jan van Haarst"
-__email__ = "jan.vanhaarst@wur.nl"
-
-
-# Download and install samtools
-rule reas_installation:
- output: CONFIG["executables"]["reas"] + 'reas_all.pl'
- params : dir = strip_path_level(CONFIG['executables']['reas'],1)
- threads : 12
- shell:
- "TEMPDIR=`mktemp -d`;"
- "cd $TEMPDIR;"
- "git clone " +CONFIG["executable_sources"]['reas_URL'] + " && "
- "cd reas && "
- "mkdir bin && "
- "cd code && "
- "make -j {threads} && "
- "make install && "
- "mv -v ../bin {params.dir} &&"
- "touch {output} && "
- "cd / && rm -rf $TEMPDIR;"
-
-
-FTP = FTPRemoteProvider()
-READS=os.path.basename(CONFIG["reads_URL"])
-
-rule download_data:
- input: FTP.remote(CONFIG["reads_URL"], keep_local=True)
- output : READS
- shell : "mv {input} {output}"
-
-
-rule reas:
- input:
- reads = "reads/{sample}/{library}/{replicate}/{name}.fastq.gz",
- output: "TOPHAT/{sample}/{library}/{replicate}/{name}/accepted_hits.bam"
- message: """Run Tophat on unpaired data"""
- threads: 8
- params:
- options = CONFIG["reas"]["options"]
- shell:
- """
- TEMPDIR=$(mktemp -d);
- UUID=$(uuidgen);
- if conda info --envs | grep '^reas'
- then
- echo "Conda environment exists"
- else
- conda create --use-index-cache --yes --name reas muscle blast;
- fi
- source activate reas
- tophat2 --num-threads {threads} {params.options} --tmp-dir $TEMPDIR --output-dir TOPHAT/{wildcards.sample}/{wildcards.library}/{wildcards.replicate}/{wildcards.name} {input.reference} {input.reads} &&
- rm -rf $TEMPDIR;
- source deactivate;
- """
+"""
+@version: 0.1
+
+Run ReAS repeat discovery tool
+
+Required programs:
+* ReAS
+* MUSCLE
+* BLASTN (from BLAST-suite)
+* dustmasker (from BLAST-suite)
+
+"""
+
+import os, subprocess
+import configparser
+from snakemake.remote.FTP import RemoteProvider as FTPRemoteProvider
+
+__author__ = "Jan van Haarst"
+__email__ = "jan.vanhaarst@wur.nl"
+
+
+# Download and install samtools
+rule reas_installation:
+ output: CONFIG["executables"]["reas"] + 'reas_all.pl'
+ params : dir = strip_path_level(CONFIG['executables']['reas'],1)
+ threads : 12
+ shell:
+ "TEMPDIR=`mktemp -d`;"
+ "cd $TEMPDIR;"
+ "git clone " +CONFIG["executable_sources"]['reas_URL'] + " && "
+ "cd reas && "
+ "mkdir bin && "
+ "cd code && "
+ "make -j {threads} && "
+ "make install && "
+ "mv -v ../bin {params.dir} &&"
+ "touch {output} && "
+ "cd / && rm -rf $TEMPDIR;"
+
+
+FTP = FTPRemoteProvider()
+READS=os.path.basename(CONFIG["reads_URL"])
+FORWARD_READS=os.path.basename(CONFIG["reads_pe_URLS"]["1"])
+REVERSE_READS=os.path.basename(CONFIG["reads_pe_URLS"]["2"])
+
+ruleorder: download_pe_data > download_data
+
+rule download_data:
+ input: FTP.remote(CONFIG["reads_URL"], keep_local=True)
+ output : READS
+ shell : "mv {input} {output}"
+
+rule download_pe_data:
+ input:
+ forward = FTP.remote(CONFIG["reads_pe_URLS"]["1"], keep_local=True),
+ reverse = FTP.remote(CONFIG["reads_pe_URLS"]["2"], keep_local=True)
+ output :
+ FORWARD_READS=FORWARD_READS,
+ REVERSE_READS=REVERSE_READS
+ shell :
+ "mv {input.forward} {output.FORWARD_READS};"
+ "mv {input.reverse} {output.REVERSE_READS};"
+
+rule reas:
+ input:
+ reads = "reads/{sample}/{library}/{replicate}/{name}.fastq.gz",
+ output: "TOPHAT/{sample}/{library}/{replicate}/{name}/accepted_hits.bam"
+ message: """Run Tophat on unpaired data"""
+ threads: 8
+ params:
+ options = CONFIG["reas"]["options"]
+ shell:
+ """
+ TEMPDIR=$(mktemp -d);
+ UUID=$(uuidgen);
+ if conda info --envs | grep '^reas'
+ then
+ echo "Conda environment exists"
+ else
+ conda create --use-index-cache --yes --name reas muscle blast;
+ fi
+ source activate reas
+ tophat2 --num-threads {threads} {params.options} --tmp-dir $TEMPDIR --output-dir TOPHAT/{wildcards.sample}/{wildcards.library}/{wildcards.replicate}/{wildcards.name} {input.reference} {input.reads} &&
+ rm -rf $TEMPDIR;
+ source deactivate;
+ """
diff --git a/src/workflows/repeats/reas/Snakefile b/src/workflows/repeats/reas/Snakefile
index 57d972f..7813ff0 100644
--- a/src/workflows/repeats/reas/Snakefile
+++ b/src/workflows/repeats/reas/Snakefile
@@ -54,5 +54,7 @@ rule all:
input:
# CONFIG['base']['bashrc'], # Needs to be separate defined always !
CONFIG["executables"]["reas"] + 'reas_all.pl',
- os.path.basename(CONFIG["reads_URL"])
+ os.path.basename(CONFIG["reads_URL"]),
+ os.path.basename(CONFIG["reads_pe_URLS"]["1"]),
+ os.path.basename(CONFIG["reads_pe_URLS"]["2"])
diff --git a/src/workflows/repeats/reas/config.json b/src/workflows/repeats/reas/config.json
index 9a8536e..5621924 100644
--- a/src/workflows/repeats/reas/config.json
+++ b/src/workflows/repeats/reas/config.json
@@ -17,23 +17,10 @@
"wget_options": " -c --proxy=off "
},
"reads_info" : "http://www.ebi.ac.uk/ena/data/view/ERX147330",
- "reads_URL" : "ftp.sra.ebi.ac.uk/vol1/fastq/ERR171/ERR171441/ERR171441_1.fastq.gz",
- "samples":
- {
- "Pleurotus_ostreatus": {
- "EP57": {
- "readsets": {
- "1": [
- "/home/lavri002/scratch/oudehome/lavri002/temp/S240_05A_CHG009446-0428lane1-EP57_L004_R1.fastq",
- "/home/lavri002/scratch/oudehome/lavri002/temp/S240_05A_CHG009446-0428lane1-EP57_L004_R2.fastq"
- ]
- },
- "type": "pe",
- "insertSize": "300",
- "insertSizeStDev": "",
- "platform": "illumina"
- }
- }
+ "reads_URL" : "ftp.sra.ebi.ac.uk/vol1/fastq/DRR001/DRR001191/DRR001191_1.fastq.gz",
+ "reads_pe_URLS" : {
+ "1": "ftp.sra.ebi.ac.uk/vol1/fastq/ERR171/ERR171441/ERR171441_1.fastq.gz",
+ "2": "ftp.sra.ebi.ac.uk/vol1/fastq/ERR171/ERR171441/ERR171441_2.fastq.gz"
},
"reas":
{
--
GitLab