From de0a927ca7205fcda289a13db5f10b7404c2936a Mon Sep 17 00:00:00 2001
From: Jan van Haarst <jan.vanhaarst@wur.nl>
Date: Fri, 18 Nov 2016 14:15:38 +0100
Subject: [PATCH] Subsample reads, so that crossmathc doesn't choke
---
src/rules/fastqProcessors/Seqtk.rules | 27 ++++++++++++++++++++++++++
src/workflows/repeats/reas/Snakefile | 3 +++
src/workflows/repeats/reas/config.json | 8 +++++++-
3 files changed, 37 insertions(+), 1 deletion(-)
create mode 100644 src/rules/fastqProcessors/Seqtk.rules
diff --git a/src/rules/fastqProcessors/Seqtk.rules b/src/rules/fastqProcessors/Seqtk.rules
new file mode 100644
index 0000000..e2c472f
--- /dev/null
+++ b/src/rules/fastqProcessors/Seqtk.rules
@@ -0,0 +1,27 @@
+"""
+@version: 0.1
+
+Use seqtk toolkit
+
+Required programs:
+* seqtk (bioconda)
+
+"""
+
+import os, subprocess
+import configparser
+
+__author__ = "Jan van Haarst"
+__email__ = "jan.vanhaarst@wur.nl"
+
+
+rule seqtk_sample:
+ input: "{name}.fasta"
+ params:
+ number = CONFIG["seqtk"]["options"]["number"],
+ seed = CONFIG["seqtk"]["options"]["seed"]
+ output: "{name}.sampled.fasta"
+ shell:
+ """
+ seqtk sample -s{params.seed} {input} {params.number} > {output}
+ """
diff --git a/src/workflows/repeats/reas/Snakefile b/src/workflows/repeats/reas/Snakefile
index 43e5b87..783063d 100644
--- a/src/workflows/repeats/reas/Snakefile
+++ b/src/workflows/repeats/reas/Snakefile
@@ -50,6 +50,9 @@ workdir: CONFIG["base"]["working_dir"]
# Include necessary rules
include: ruleDir + "/repeats/reas.rules"
include: ruleDir + "/fastqProcessors/Sra.rules"
+include: ruleDir + "/fastqProcessors/Seqtk.rules"
+
+os.mkdir(strip_path_level(CONFIG['executables']['reas'],1))
rule all:
input:
diff --git a/src/workflows/repeats/reas/config.json b/src/workflows/repeats/reas/config.json
index 16de7c2..9395c12 100644
--- a/src/workflows/repeats/reas/config.json
+++ b/src/workflows/repeats/reas/config.json
@@ -48,6 +48,12 @@
"seg_file" : "seg.fa"
}
},
+ "seqtk":{
+ "options":{
+ "number" : "1000000",
+ "seed" : "314159"
+ }
+ },
"sra":
{
"sra_download" :{
@@ -57,5 +63,5 @@
"options" : "--split-files --fasta"
}
},
- "reads" : "{download_dir}sra/ERR171441_1.fasta"
+ "reads" : "{download_dir}sra/ERR171441_1.sampled.fasta"
}
--
GitLab