From 710a67d9f766777667679e24af93212ea764a928 Mon Sep 17 00:00:00 2001
From: "Gervaise H. Henry" <gervaise.henry@utsouthwestern.edu>
Date: Sat, 18 Jan 2020 17:55:28 -0600
Subject: [PATCH] Add checks for meta repRID
---
workflow/conf/aws_ondemand.config | 7 -------
workflow/conf/aws_spot.config | 7 -------
workflow/conf/biohpc.config | 6 +++---
workflow/nextflow.config | 15 +++++++++++++++
workflow/rna-seq.nf | 1 +
workflow/scripts/parseMeta.py | 17 +++++++++++------
6 files changed, 30 insertions(+), 23 deletions(-)
diff --git a/workflow/conf/aws_ondemand.config b/workflow/conf/aws_ondemand.config
index 1a14ebf..84fcb27 100755
--- a/workflow/conf/aws_ondemand.config
+++ b/workflow/conf/aws_ondemand.config
@@ -13,14 +13,7 @@ process {
cpus = 1
memory = '1 GB'
- withName:getBag {
- container = 'bicf/gudmaprbkfilexfer:1.3'
- }
- withName:getData {
- container = 'bicf/gudmaprbkfilexfer:1.3'
- }
withName:trimData {
- container = 'bicf/trimgalore:1.1'
cpus = 15
}
}
\ No newline at end of file
diff --git a/workflow/conf/aws_spot.config b/workflow/conf/aws_spot.config
index b5239a2..fbccb3c 100755
--- a/workflow/conf/aws_spot.config
+++ b/workflow/conf/aws_spot.config
@@ -13,14 +13,7 @@ process {
cpus = 1
memory = '1 GB'
- withName:getBag {
- container = 'bicf/gudmaprbkfilexfer:1.3'
- }
- withName:getData {
- container = 'bicf/gudmaprbkfilexfer:1.3'
- }
withName:trimData {
- container = 'bicf/trimgalore:1.1'
cpus = 15
}
}
diff --git a/workflow/conf/biohpc.config b/workflow/conf/biohpc.config
index 20da91a..36d5b33 100755
--- a/workflow/conf/biohpc.config
+++ b/workflow/conf/biohpc.config
@@ -5,16 +5,16 @@ process {
withName:getBag {
executor = 'local'
- container = 'docker://bicf/gudmaprbkfilexfer:1.3'
}
withName:getData {
executor = 'local'
- container = 'docker://bicf/gudmaprbkfilexfer:1.3'
}
withName:trimData {
- container = 'docker://bicf/trimgalore:1.1'
queue = '256GB,256GBv1,384GB'
}
+ withName:parseMetadata {
+ executor = 'local'
+ }
}
singularity {
diff --git a/workflow/nextflow.config b/workflow/nextflow.config
index 3758499..2635a79 100644
--- a/workflow/nextflow.config
+++ b/workflow/nextflow.config
@@ -10,6 +10,21 @@ profiles {
}
}
+process {
+ withName:getBag {
+ container = 'bicf/gudmaprbkfilexfer:1.3'
+ }
+ withName:getData {
+ container = 'bicf/gudmaprbkfilexfer:1.3'
+ }
+ withName:trimData {
+ container = 'bicf/trimgalore:1.1'
+ }
+ withName:parseMetadata {
+ container = 'bicf/python:1.3'
+ }
+}
+
trace {
enabled = true
file = 'pipeline_trace.txt'
diff --git a/workflow/rna-seq.nf b/workflow/rna-seq.nf
index 6e4199d..a10a29a 100755
--- a/workflow/rna-seq.nf
+++ b/workflow/rna-seq.nf
@@ -111,6 +111,7 @@ process parseMetadata {
"""
hostname >>${repRID_parseMetadata}.parseMetadata.err
ulimit -a >>${repRID_parseMetadata}.parseMetadata.err
+ python ${baseDir}/scripts/parseMeta.py -r ${repRID_parseMetadata} -m ${fileMeta} -p repRID
"""
}
diff --git a/workflow/scripts/parseMeta.py b/workflow/scripts/parseMeta.py
index 115dd6d..bd872fc 100644
--- a/workflow/scripts/parseMeta.py
+++ b/workflow/scripts/parseMeta.py
@@ -14,14 +14,19 @@ def get_args():
def main():
args = get_args()
- metaFile = pd.read_csv(args.metaFile,sep="\t",header=None)
+ metaFile = pd.read_csv(args.metaFile,sep=",",header=0)
if (args.parameter == "repRID"):
if (len(metaFile.Replicate_RID.unique()) > 1):
- #ERROR
- if not (metaFile.Replicate_RID == arg$repRID):
- #ERROR
- if (len(fileFile[fileFile["File_Type"] == "FastQ"].RID) > 2):
- #ERROR
+ print("There are multiple replicate RID's in the metadata: " + " ".join(metaFile.Replicate_RID.unique()))
+ exit(1)
+ if not (metaFile.Replicate_RID.unique() == args.repRID):
+ print("Replicate RID in metadata does not match run parameters: " + metaFile.Replicate_RID.unique() + " vs " + args.repRID)
+ exit(1)
+ else:
+ print(metaFile["Replicate_RID"].unique())
+ if (len(metaFile[metaFile["File_Type"] == "FastQ"]) > 2):
+ print("There are more then 2 fastq's in the metadata: " + " ".join(metaFile[metaFile["File_Type"] == "FastQ"].RID))
+ exit(1)
if __name__ == '__main__':
--
GitLab