def loadFastqc(infile, outfile):
'''load FASTQC stats into database.'''
track = P.snip(infile, ".fastqc")
filename = os.path.join(PARAMS["exportdir"], "fastqc", track + "*_fastqc",
"fastqc_data.txt")
PipelineReadqc.loadFastqc(filename, database_url=PARAMS["database"]["url"])
IOTools.touch_file(outfile)
def buildExperimentLevelReadQuality(infiles, outfile):
"""Collate per sequence read qualities for all replicates per
experiment. Replicates are the last part of a filename,
eg. Experiment-R1, Experiment-R2, etc.
"""
PipelineReadqc.buildExperimentReadQuality(infiles, outfile, "fastqc.dir")
def buildExperimentLevelReadQuality(infiles, outfile):
"""
Collate per sequence read qualities for all replicates per experiment.
Replicates are the last part of a filename, eg. Experiment-R1,
Experiment-R2, etc.
"""
exportdir = os.path.join(PARAMS["exportdir"], "fastqc")
PipelineReadqc.buildExperimentReadQuality(infiles, outfile, exportdir)
def buildExperimentLevelReadQuality(infiles, outfile):
"""
Collate per sequence read qualities for all replicates per experiment.
Replicates are the last part of a filename, eg. Experiment-R1,
Experiment-R2, etc.
"""
exportdir = os.path.join(PARAMS["exportdir"], "fastqc")
PipelineReadqc.buildExperimentReadQuality(infiles, outfile, exportdir)
def loadFastqc(infile, outfile):
'''load FASTQC stats into database.'''
track = P.snip(infile, ".fastqc")
filename = os.path.join(
PARAMS["exportdir"], "fastqc", track + "*_fastqc", "fastqc_data.txt")
PipelineReadqc.loadFastqc(filename,
backend=PARAMS["database_backend"],
database=PARAMS["database_name"],
host=PARAMS["database_host"],
username=PARAMS["database_username"],
password=PARAMS["database_password"],
port=PARAMS["database_port"])
P.touch(outfile)
def loadFastqc(infile, outfile):
'''load FASTQC stats into database.'''
track = P.snip(infile.replace("processed.dir/", ""), ".fastqc")
filename = os.path.join(
PARAMS["exportdir"], "fastqc", track + "*_fastqc", "fastqc_data.txt")
PipelineReadqc.loadFastqc(filename,
backend=PARAMS["database_backend"],
database=PARAMS["database_name"],
host=PARAMS["database_host"],
username=PARAMS["database_username"],
password=PARAMS["database_password"],
port=PARAMS["database_port"])
P.touch(outfile)
def summarizeFastqScreen(infiles, outfiles):
all_files = []
for infile in infiles:
all_files.extend(glob.glob(IOTools.snip(infile, "screen") + "*_screen.txt"))
if len(all_files) == 0:
E.warn("no fastqcscreen results to concatenate")
for x in outfiles:
IOTools.touch_file(x)
return
df_summary, df_details = PipelineReadqc.read_fastq_screen(
all_files)
df_summary.to_csv(outfiles[0], sep="\t", index=True)
df_details.to_csv(outfiles[1], sep="\t", index=True)
def summarizeFastQC(infiles, outfiles):
all_files = []
for infile in infiles:
track = P.snip(infile, ".fastqc")
all_files.extend(
glob.glob(os.path.join(track + "*_fastqc", "fastqc_data.txt")))
dfs = PipelineReadqc.read_fastqc(all_files)
for key, df in dfs.items():
fn = re.sub("basic_statistics", key, outfiles[0])
E.info("writing to {}".format(fn))
with IOTools.open_file(fn, "w") as outf:
df.to_csv(outf, sep="\t", index=True)
def buildFastQCSummaryBasicStatistics(infiles, outfile):
'''load fastqc summaries into a single table.'''
exportdir = os.path.join(PARAMS["exportdir"], "fastqc")
PipelineReadqc.buildFastQCSummaryBasicStatistics(infiles, outfile,
exportdir)
def buildFastQCSummaryBasicStatistics(infiles, outfile):
'''load fastqc summaries into a single table.'''
exportdir = os.path.join(PARAMS["exportdir"], "fastqc")
PipelineReadqc.buildFastQCSummaryBasicStatistics(infiles, outfile,
exportdir)
def buildFastQCSummaryStatus(infiles, outfile):
'''load FastQC status summaries into a single table.'''
PipelineReadqc.buildFastQCSummaryStatus(infiles, outfile, "fastqc.dir")
def buildFastQCSummaryStatus(infiles, outfile):
"""load fastqc status summaries into a single table."""
exportdir = os.path.join(PARAMS["exportdir"], "fastqc")
PipelineReadqc.buildFastQCSummaryStatus(infiles, outfile, exportdir)
