1
1
#! / usr/ bin/ env nextflow
2
2
nextflow. enable. dsl = 2
3
3
4
- // params.GLDS_accession = "OSD-574"
4
+ // params.accession = "OSD-574"
5
5
// params.RawFilePattern = null // Pattern of files on OSDR for the OSD accession you want to process
6
6
7
7
process GET_RUNSHEET {
8
8
9
9
beforeScript " chmod +x ${ baseDir} /bin/create_runsheet.sh"
10
10
11
11
input:
12
- val(GLDS_accession )
12
+ val(accession )
13
13
output:
14
14
path(" a_*metagenomic*.txt" ), emit: assay_TABLE
15
15
path(" *.zip" ), emit: zip
@@ -18,19 +18,19 @@ process GET_RUNSHEET {
18
18
script:
19
19
"""
20
20
# Download ISA zip file for the GLDS_accession then unzip it
21
- GL-download-GLDS-data -g ${ GLDS_accession } -p ISA -f && unzip *-ISA.zip
21
+ GL-download-GLDS-data -g ${ accession } -p ISA -f && unzip *-ISA.zip
22
22
23
23
if [ ${ params.RawFilePattern} == null ];then
24
24
25
25
# Attempt to download the sequences using the assay table, if that fails then
26
26
# attempt retrieving all fastq.gz files
27
- GL-download-GLDS-data -f -g ${ GLDS_accession } -a a_*metagenomic*.txt -o Raw_Sequence_Data || \\
28
- GL-download-GLDS-data -f -g ${ GLDS_accession } -p ".fastq.gz" -o Raw_Sequence_Data
27
+ GL-download-GLDS-data -f -g ${ accession } -a a_*metagenomic*.txt -o Raw_Sequence_Data || \\
28
+ GL-download-GLDS-data -f -g ${ accession } -p ".fastq.gz" -o Raw_Sequence_Data
29
29
30
30
else
31
31
32
32
33
- GL-download-GLDS-data -f -g ${ GLDS_accession } -p ${ params.RawFilePattern} -o Raw_Sequence_Data
33
+ GL-download-GLDS-data -f -g ${ accession } -p ${ params.RawFilePattern} -o Raw_Sequence_Data
34
34
35
35
fi
36
36
@@ -39,8 +39,8 @@ process GET_RUNSHEET {
39
39
grep '+' *wanted-file-download-commands.sh | \\
40
40
sort -u | \\
41
41
awk '{gsub(/\\ +/,"%2B", \$ NF);print}' \\
42
- > plus_containing_${ GLDS_accession } -wanted-file-download-commands.sh
43
- cat plus_containing_${ GLDS_accession } -wanted-file-download-commands.sh | parallel -j $task . cpus
42
+ > plus_containing_${ accession } -wanted-file-download-commands.sh
43
+ cat plus_containing_${ accession } -wanted-file-download-commands.sh | parallel -j $task . cpus
44
44
fi
45
45
46
46
# Create runsheet from the assay table
@@ -52,7 +52,7 @@ process GET_RUNSHEET {
52
52
53
53
workflow {
54
54
55
- GET_RUNSHEET (params.GLDS_accession )
55
+ GET_RUNSHEET (params. accession )
56
56
file_ch = GET_RUNSHEET . out. input_file
57
57
.splitCsv(header :true )
58
58
0 commit comments