Skip to content

Commit

Permalink
fix f-string issue, add log
Browse files Browse the repository at this point in the history
  • Loading branch information
@alienzj
alienzj committed Aug 18, 2022
1 parent 591f9e6 commit 2e5bcec
Showing 1 changed file with 44 additions and 43 deletions.
87 changes: 44 additions & 43 deletions metapi/rules/raw.smk
View file
Original file line number Diff line number Diff line change
Expand Up @@ -81,77 +81,78 @@ if config["params"]["raw"]["do"]:
os.symlink(os.path.realpath(input[0]), f'''{output.reads[0]}.temp.gz''')
os.symlink(os.path.realpath(input[1]), f'''{output.reads[1]}.temp.gz''')
else:
shell(f'''cat {" ".join(input[0:reads_num//2])} > {output.reads[0]}.temp.gz''')
shell(f'''cat {" ".join(input[reads_num//2:])} > {output.reads[1]}.temp.gz''')
shell(f'''cat {" ".join(input[0:reads_num//2])} > {output.reads[0]}.temp.gz 2>> {log}''')
shell(f'''cat {" ".join(input[reads_num//2:])} > {output.reads[1]}.temp.gz 2>> {log}''')

shell(f'''seqkit seq -ni {output.reads[0]}.temp.gz | sed 's#/1$##g' > {params.output_dir}/id.list.1''')
shell(f'''seqkit seq -ni {output.reads[1]}.temp.gz | sed 's#/2$##g' > {params.output_dir}/id.list.2''')
shell(f'''seqkit seq -ni {output.reads[0]}.temp.gz | sed 's#/1$##g' > {params.output_dir}/id.list.1 2>> {log}''')
shell(f'''seqkit seq -ni {output.reads[1]}.temp.gz | sed 's#/2$##g' > {params.output_dir}/id.list.2 2>> {log}''')

if filecmp.cmp(f'''{params.output_dir}/id.list.1''', f'''{params.output_dir}/id.list.2'''):
shell(f'''mv {output.reads[0]}.temp.gz {output.reads[0]}''')
shell(f'''mv {output.reads[1]}.temp.gz {output.reads[1]}''')
shell(f'''mv {output.reads[0]}.temp.gz {output.reads[0]} 2>> {log}''')
shell(f'''mv {output.reads[1]}.temp.gz {output.reads[1]} 2>> {log}''')
else:
shell(
f'''
cat {params.output_dir}/id.list.1 {params.output_dir}/id.list.2 | \
sort -T {params.output_dir} | \
awk '$1==2{{print $2}}' > {params.output_dir}/id.list.paired
''')
'''
sort -T {params.output_dir} {params.output_dir}/id.list.1 {params.output_dir}/id.list.2 | \
uniq -c | \
awk '$1==2{{print $2}}' > {params.output_dir}/id.list.paired 2>> {log}
''')

oneline = gzip.open(f'''{output.reads[0]}.temp.gz''', 'rt').readline().strip().split()[0]
if "/1" in oneline:
shell(
f'''
seqkit grep -f <(awk '{{print $0 "/1"}}' {params.output_dir}/id.list.paired) {output.reads[0]}.temp.gz -o {output.reads[0]}
seqkit grep -f <(awk '{{print $0 "/2"}}' {params.output_dir}/id.list.paired) {output.reads[1]}.temp.gz -o {output.reads[1]}
'''
seqkit grep -f <(awk '{{print $0 "/1"}}' {params.output_dir}/id.list.paired) {output.reads[0]}.temp.gz -o {output.reads[0]} 2>> {log}
seqkit grep -f <(awk '{{print $0 "/2"}}' {params.output_dir}/id.list.paired) {output.reads[1]}.temp.gz -o {output.reads[1]} 2>> {log}
''')
else:
shell(
f'''
seqkit grep -f {params.output_dir}/id.list.paired {output.reads[0]}.temp.gz -o {output.reads[0]}
seqkit grep -f {params.output_dir}/id.list.paired {output.reads[1]}.temp.gz -o {output.reads[1]}
seqkit grep -f {params.output_dir}/id.list.paired {output.reads[0]}.temp.gz -o {output.reads[0]} 2>> {log}
seqkit grep -f {params.output_dir}/id.list.paired {output.reads[1]}.temp.gz -o {output.reads[1]} 2>> {log}
''')

shell(f'''rm -rf {output.reads[0]}.temp.gz''')
shell(f'''rm -rf {output.reads[1]}.temp.gz''')
shell(f'''rm -rf {params.output_dir}/id.list.paired''')
shell(f'''rm -rf {output.reads[0]}.temp.gz 2>> {log}''')
shell(f'''rm -rf {output.reads[1]}.temp.gz 2>> {log}''')
shell(f'''rm -rf {params.output_dir}/id.list.paired 2>> {log}''')

shell(f'''rm -rf {params.output_dir}/id.list.1''')
shell(f'''rm -rf {params.output_dir}/id.list.2''')
shell(f'''rm -rf {params.output_dir}/id.list.1 2>> {log}''')
shell(f'''rm -rf {params.output_dir}/id.list.2 2>> {log}''')

else:
shell(
'''
cat {input} | \
tee >(seqtk seq -1 - | pigz -c -p {threads} > {output.reads[0]}) | \
seqtk seq -2 - | pigz -c -p {threads} > {output.reads[1]}
seqtk seq -2 - | pigz -c -p {threads} > {output.reads[1]} 2>> {log}
''')
else:
if reads_num == 1:
os.symlink(os.path.realpath(input[0]), output.reads[0])
else:
shell('''cat {input} > {output.reads[0]}''')
shell('''cat {input} > {output.reads[0]} 2>> {log}''')

elif READS_FORMAT == "sra":
if reads_num == 1:
sra_file = os.path.basename(input[0])
shell(
f'''
rm -rf {params.output_dir}/{sra_file}*
rm -rf {params.output_dir}.{sra_file}.temp
rm -rf {params.output_dir}/{sra_file}* 2>> {log}
rm -rf {params.output_dir}.{sra_file}.temp 2>> {log}
fasterq-dump \
--threads {threads} \
--split-3 \
--temp {params.output_dir}.{sra_file}.temp \
--outdir {params.output_dir} {input[0]} >{log} 2>&1
--outdir {params.output_dir} {input[0]} 2>>{log}
rm -rf {params.output_dir}.{sra_file}.temp
pigz --processes {threads} {params.output_dir}/{sra_file}_1.fastq
pigz --processes {threads} {params.output_dir}/{sra_file}_2.fastq
rm -rf {params.output_dir}/{sra_file}._*.fastq
rm -rf {params.output_dir}.{sra_file}.temp 2>> {log}
pigz --processes {threads} {params.output_dir}/{sra_file}_1.fastq 2>> {log}
pigz --processes {threads} {params.output_dir}/{sra_file}_2.fastq 2>> {log}
rm -rf {params.output_dir}/{sra_file}._*.fastq 2>> {log}
mv {params.output_dir}/{sra_file}_1.fastq.gz {output.reads[0]}
mv {params.output_dir}/{sra_file}_2.fastq.gz {output.reads[1]}
mv {params.output_dir}/{sra_file}_1.fastq.gz {output.reads[0]} 2>> {log}
mv {params.output_dir}/{sra_file}_2.fastq.gz {output.reads[1]} 2>> {log}
''')

else:
Expand All @@ -165,27 +166,27 @@ if config["params"]["raw"]["do"]:
sra_file + "_2.fastq.gz"))
shell(
f'''
rm -rf {params.output_dir}/{sra_file}*
rm -rf {params.output_dir}.{sra_file}.temp
rm -rf {params.output_dir}/{sra_file}* 2>> {log}
rm -rf {params.output_dir}.{sra_file}.temp 2>> {log}
fasterq-dump \
--threads {threads} \
--split-3 \
--temp {params.output_dir}.{sra_file}.temp \
--outdir {params.output_dir} {sra} >>{log} 2>&1
--outdir {params.output_dir} {sra} 2>> {log}
rm -rf {params.output_dir}.{sra_file}.temp
pigz --processes {threads} {params.output_dir}/{sra_file}_1.fastq
pigz --processes {threads} {params.output_dir}/{sra_file}_2.fastq
rm -rf {params.output_dir}/{sra_file}._*.fastq
rm -rf {params.output_dir}.{sra_file}.temp 2>> {log}
pigz --processes {threads} {params.output_dir}/{sra_file}_1.fastq 2>> {log}
pigz --processes {threads} {params.output_dir}/{sra_file}_2.fastq 2>> {log}
rm -rf {params.output_dir}/{sra_file}._*.fastq 2>> {log}
''')

r1_str = " ".join(r1_list)
r2_str = " ".join(r2_list)
shell('''cat %s > %s''' % (r1_str, output.reads[0]))
shell('''cat %s > %s''' % (r2_str, output.reads[1]))
shell('''rm -rf %s''' % r1_str)
shell('''rm -rf %s''' % r2_str)
shell(f'''cat {r1_str} > {output.reads[0]} 2>> {log}''')
shell(f'''cat {r2_str} > {output.reads[1]} 2>> {log}''')
shell(f'''rm -rf {r1_str} 2>> {log}''')
shell(f'''rm -rf {r2_str} 2>> {log}''')


rule prepare_short_reads_all:
Expand Down

0 comments on commit 2e5bcec

Please sign in to comment.