def _func(job, key, WORKDIR, chipseq_targets_peaks):
fig, axs = plt.subplots(2, 2, figsize=[12, 12])
axs = axs.ravel()
with pyext.getPathStack([WORKDIR, key], force=1):
GSIZE = "/home/feng/ref/ATH-TAIR10/genome.sizes"
peakFile = chipseq_targets_peaks['LAST_FILE']
featFile = sutil.bed__leftSummit(
"/home/feng/ref/ATH-TAIR10/annotation/genes.gtf.cds", GSIZE=GSIZE)
pyext.file__link(peakFile, "PEAK_FILE.bed", force=1)
pyext.file__link(featFile, "FEAT_FILE.bed", force=1)
res = synotil.qcplots.qc_summitDist(
# chipseq_targets_genes_peaks,
peakFile,
featFile,
GSIZE=GSIZE,
axs=axs,
CUTOFF=500,
)
df = res[0]
# pyext.readData(peakFile,header=None).set_index(3,drop=0).reindex(df['acc'])
# df = df.set_index('acc',drop=0).reindex(pyext.readData(peakFile,header=None).set_index(3))
df.to_csv("PEAK_DIST.csv", index=0)
lst = df.dropna()['feat_acc'].drop_duplicates().values.tolist()
with pyext.getPathStack([WORKDIR, key], force=1) as stack:
pyext.printlines(lst, "OUT.it")
return dict(LAST_DIR=WORKDIR / key)
def get_soft_text():
# OUTPUT_FILE = WORKDIR() / _fkey() ("get_soft_text.tar.gz")
# dfc = meta_df_rnaseq()
BUFFER_ALL = BUFFER_RNASEQ() + BUFFER_CHIPSEQ()
FORCE = 1
OUTDIR = WORKDIR()/_fkey()
shell.real__dir(dirname= OUTDIR)
if not FORCE and shell.file__notEmpty(OUTDIR + '.tar.gz'):
pass
else:
df_figureRegistry().to_csv(OUTDIR / 'figureRegistry.csv',index=0)
def _worker(sample):
sample = attrdict.AttrDict(sample)
sample.title = "_".join([sample[k] for k in "data_acc,age,tissue,genotype,ztime,temperature".split(",")])
res = res= pyext.jf2( template_common())
res = '\n'.join([x.strip() for x in res.splitlines()])
sample.soft_text = res
pyext.printlines([sample.soft_text], OUTDIR / pyext.f("{sample.data_acc}.soft.txt"))
# %sample.data_acc)
[_worker(sample) for sample in pyext.df__iterdict( meta_df_rnaseq().fillna("NA"))]
[_worker(sample) for sample in pyext.df__iterdict( meta_df_chipseq().fillna("NA"))]
with pyext.TempDirScope(getTempDirName=lambda: OUTDIR/'_temp',
force=1) as tdir:
tdir = tdir.d
# tdir = OUTDIR / '_temp'
shell.real__dir(dirname= tdir)
# shell.shellexec(' '.join(["tar" ,"-C",tdir,"-xvzf", INPUTDIR()/"./src/polyq-0830.get_soft_text_jaehoon.tar.gz",]))
# shell.shellexec(' '.join(["tar" ,"-C",tdir,"-xvzf", INPUTDIR()/"./src/polyq-0905.get_soft_text_Jaehoon0906.tar.gz",]))
# shell.shellexec(' '.join(["tar" ,"-C",tdir,"-xvzf", INPUTDIR()/"./src/polyq-0907.katja.get_soft_text.tar.gz",]))
shell.shellexec(' '.join(["unzip","-d",tdir, "-o",INPUTDIR()/"./src/polyq-0907-jaehoon.get_soft_text.zip", ]))
for fname in tdir.glob('*.soft.txt'):
DATA_ACC = fname.basename().replace('.soft.txt','',1)
# in df_figureRegistry()['DATA_ACC'].tolist()
assert DATA_ACC in df_figureRegistry()['DATA_ACC'].tolist(),(DATA_ACC, "JH0905")
print('[JH0906]',fname.basename(),)
pyext.file__link(fname, OUTDIR/fname.basename(),force=1)
CMD = [
"cd",OUTDIR,
"&&tar","-cvzf",OUTDIR.realpath()+'.tar.gz',"*",
]
CMD = ' '.join(pyext.stringList__flatten(CMD))
res = shell.shellexec(CMD)
"cd",
OUTDIR,
"&&cat",
"SERIES.soft.txt",
"*autofill*",
"|grep",
"-v",
"^#",
"|grep",
"-v",
"^$",
"|sed",
u"'s/^\xEF\xBB\xBF//g'",
">" + OUTDIR / OFNAME,
# "|tee","0830-polyq-submit.soft",
]),
silent=1)
pyext.file__link(OUTDIR / OFNAME, WORKDIR() / "ftp" / OFNAME, force=1)
CMD = [
"cd",
OUTDIR,
"&&tar",
"-cvzf",
OUTDIR.realpath() + '.tar.gz',
"*",
]
CMD = ' '.join(pyext.stringList__flatten(CMD))
res = shell.shellexec(CMD)
# pyext.printlines([pyext.jf2(template)], WORKDIR()/ 'get_soft_text' / '0829-polyq.SERIES.soft.txt')
# break