From eced405aa0cc351fdfa1cce42d598a79c6e10e00 Mon Sep 17 00:00:00 2001
From: nikos <n.pappas@uu.nl>
Date: Mon, 11 Jan 2021 13:43:25 +0100
Subject: [PATCH] change samplesheet to tsv
---
workflow/Snakefile | 23 ++++++++++++++---------
1 file changed, 14 insertions(+), 9 deletions(-)
diff --git a/workflow/Snakefile b/workflow/Snakefile
index f35d93d..bb1583b 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -2,18 +2,23 @@ from pathlib import Path
configfile: "config/config.yaml"
+# This doesn't work for now
+# Need to bind mount the data directory on the containers
+# with --singularity-args "-B path/to/data_dir:/data"
DATA_DIR = Path(config.get("vhmnet").get("data_dir"))
-def parse_samples_csv(samples_csv):
- samples = {}
- with open(samples_csv, 'r') as fin:
- next(fin)
+def parse_samplesheet(samples_tsv):
+ samples_dic = {}
+ with open(samples_tsv, 'r') as fin:
+ header_line = fin.readline()
+ header_fields = [f.strip() for f in header_line.split('\t')]
+ assert header_fields == ['sample', 'fasta'], "Malformatted samplesheet"
for line in fin:
- fields = [f.strip() for f in line.split(',')]
- samples[fields[0]] = fields[1]
- return samples
+ fields = [f.strip() for f in line.split('\t')]
+ samples_dic[fields[0]] = fields[1]
+ return samples_dic
-samples_dic = parse_samples_csv(config.get('samplesheet', 'samples.csv'))
+samples_dic = parse_samplesheet(config.get('samplesheet', 'samples.tsv'))
SAMPLES = list(samples_dic.keys())
TOOLS = [
@@ -117,6 +122,7 @@ rule filter_rafah:
shell:
"tail -n+2 {input.seq_info} | cut -f1,6,7 | sort -k1 "
"> {output.rafah_tsv}"
+
# VirHostMatcher-Net
rule run_vhmnet:
input:
@@ -158,7 +164,6 @@ rule filter_vhmnet:
"""
for f in $(find -wholename "{params.predictions_dir}/*.csv" -type f);
do
- echo ${{f}};
contig_id=$(basename ${{f}} | sed -e 's/_prediction.csv//')
host_score=$(tail -n1 ${{f}} | cut -f8,10 -d',' | tr ',' '\t')
echo -e "$contig_id\t$host_score" >> {output.vhmnet_tsv}.tmp;
--
GitLab