def test_pacbio_random():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
with TempFile() as fh:
b.random_selection(fh.name, nreads=10)
with TempFile() as fh:
b.random_selection(fh.name, expected_coverage=10,
reference_length=10000)
def test_pacbio_random():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
with TempFile() as fh:
b.random_selection(fh.name, nreads=10)
with TempFile() as fh:
b.random_selection(fh.name, expected_coverage=10,
reference_length=10000)
def __init__(self, filename):
self.filename = filename
self.bam = PacbioSubreads(self.filename)
self._df = None
class PacbioIsoSeqMappedIsoforms(object):
"""Here, we load a SAM/BAM file generated with minimap using
as input the BAM file created with the mapping og HQ isoforms
on a reference.
df contains a dataframe for each read found in the SAM (and hq_isoform)
we populate the GC content, the mapping flag, the reference name (-1 means
no mapping i.e flag ==4). flag of 4 means unmapped and there is no
ambiguity about it.
In the data file example, other falgs are 0, 16 (SEQ being reverse
complement<F12>) , 2048 (supplementary segment).
Example of minimap2 command::
minimap2 -t 4 -ax splice -uf --secondary=no SIRV-E0.fa
hq_isoforms.fasta 1> hq_isoforms.fasta.sam 2> hq_isoforms.fasta.sam.log
Reads a SAM file for now. BAM should work as well
"""
def __init__(self, filename):
self.filename = filename
self.bam = PacbioSubreads(self.filename)
self._df = None
@property
def df(self):
if self._df is not None:
return self._df
# !! for isoseq, we should be able to load everything into memory
self.bam.reset()
data = [a for a in self.bam.data]
df = self.bam.df.copy()
rnames = [self.bam.data.get_reference_name(a.rname) if a.rname!=-1
else -1 for a in data]
df['reference_name'] = rnames
df['flags'] = [a.flag for a in data]
df['mapq'] = [a.mapq for a in data]
df['cigar'] = [a.cigarstring for a in data]
df['qname'] = [a.qname for a in data]
# Drop SNR that are not populated in the mapped BAM file.
df.drop(['snr_A', 'snr_C', 'snr_G', 'snr_T'], axis=1, inplace=True)
# TODO. input could be basde on mapping of CCS in which case, the ZMW is
# stored and the following does not work. could check whether the
# pattern is pXXfXX
try:
df["full_length"] = df["qname"].apply(lambda x: int(x.split('/')[1].split("p")[0].strip("f")))
df["non_full_length"] = df["qname"].apply(lambda x: int(x.split("/")[1].split("p")[1].strip("f")))
except:
pass
self._df = df
return self._df
def hist_isoform_length_mapped_vs_unmapped(self, bins=None):
df = self.df
if bins is None:
bins = range(0, len(df.reference_length.max()), 100)
mapped = df[df.reference_name != -1]
unmapped = df[df.reference_name == -1]
pylab.hist(mapped.reference_length, bins=bins, alpha=0.5,
label="mapped {}".format(len(mapped)), density=False)
pylab.hist(unmapped.reference, bins=bins, alpha=0.5,
label="unmapped {}".format(len(unmapped)), density=False)
pylab.xlabel("Isoform length")
pylab.legend()
def hist_transcript(self, hide_unmapped=True):
pylab.clf()
if hide_unmapped is True:
query = "reference_length>0 and reference_name!=-1"
else:
query = "reference_length>0"
print(query)
ts = self.df.query(query).groupby("reference_name").count().reference_length
if len(ts) == 0:
print("nothing to plot")
return ts
ts.plot(kind="bar" ,color="r")
try: pylab.tight_layout()
except: pass
return ts
def bar_mapq(self, logy=True, xmin=0, xmax=60, fontsize=12):
self.df.mapq.hist()
if logy:
pylab.semilogy()
pylab.xlim([xmin, xmax])
pylab.xlabel("Mapping quality", fontsize=fontsize)
def plot_sirv_by_group(self, title, shift=5, plot=False, mapq_min=-1):
aa = self.df.query("reference_name not in [-1, '-1']").copy()
if len(aa) == 0:
return pd.Series(), self.df
aa['group'] = aa.reference_name.apply(lambda x: x[0:shift])
mapped = aa.query("mapq>@mapq_min").groupby("group").count()["mapq"]
mapped.name = None
if plot:
mapped.plot(kind="bar")
pylab.title(title)
pylab.tight_layout()
#data.to_csv(path + "_hq_sirv_grouped.csv")
return mapped, self.df
def test_pacbio_stride():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
with TempFile() as fh:
b.stride(fh.name, stride=2)
with TempFile() as fh:
b.stride(fh.name, stride=2, random=True)
def test_pacbio():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
assert len(b) == 130
b.df
#assert b.nb_pass[1] == 130
with TempFile() as fh:
b.filter_length(fh.name, threshold_min=500)
print(b) # check length
assert b.stats['mean_GC'] > 62.46
assert b.stats['mean_GC'] < 65.47
b.summary()
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
# test hist_snr from scratch
b._df = None
b.hist_snr()
# test hist_len from scratch
b._df = None
b.hist_read_length()
b.hist_nb_passes()
b.get_mean_nb_passes()
# test from scratch
b._df = None
b.hist_GC()
# test from scratch
b._df = None
b.plot_GC_read_len()
# test from scratch
b._df = None
with TempFile() as fh:
b.to_fasta(fh.name, threads=1)
with TempFile() as fh:
b.to_fastq(fh.name, threads=1)
with TempFile() as fh:
b.save_summary(fh.name)
def __init__(self, filename):
self.filename = filename
self.bam = PacbioSubreads(self.filename)
self._df = None
class PacbioIsoSeqMappedIsoforms(object):
"""Here, we load a SAM/BAM file generated with minimap using
as input the BAM file created with the mapping og HQ isoforms
on a reference.
df contains a dataframe for each read found in the SAM (and hq_isoform)
we populate the GC content, the mapping flag, the reference name (-1 means
no mapping i.e flag ==4). flag of 4 means unmapped and there is no
ambiguity about it.
In the data file example, other flags are 0, 16 (SEQ being reverse
complement<F12>) , 2048 (supplementary segment).
Example of minimap2 command::
minimap2 -t 4 -ax splice -uf --secondary=no SIRV-E0.fa
hq_isoforms.fasta 1> hq_isoforms.fasta.sam 2> hq_isoforms.fasta.sam.log
Reads a SAM file for now. BAM should work as well
"""
def __init__(self, filename):
self.filename = filename
self.bam = PacbioSubreads(self.filename)
self._df = None
@property
def df(self):
if self._df is not None:
return self._df
# !! for isoseq, we should be able to load everything into memory
self.bam.reset()
data = [a for a in self.bam.data]
df = self.bam.df.copy()
rnames = [
self.bam.data.get_reference_name(a.rname) if a.rname != -1 else -1
for a in data
]
df['reference_name'] = rnames
df['flags'] = [a.flag for a in data]
df['mapq'] = [a.mapq for a in data]
df['cigar'] = [a.cigarstring for a in data]
df['qname'] = [a.qname for a in data]
# Drop SNR that are not populated in the mapped BAM file.
df.drop(['snr_A', 'snr_C', 'snr_G', 'snr_T'], axis=1, inplace=True)
# TODO. input could be basde on mapping of CCS in which case, the ZMW is
# stored and the following does not work. could check whether the
# pattern is pXXfXX
try:
df["full_length"] = df["qname"].apply(
lambda x: int(x.split('/')[1].split("p")[0].strip("f")))
df["non_full_length"] = df["qname"].apply(
lambda x: int(x.split("/")[1].split("p")[1].strip("f")))
except:
pass
self._df = df
return self._df
def hist_isoform_length_mapped_vs_unmapped(self, bins=None):
df = self.df
if bins is None:
bins = range(0, len(df.reference_length.max()), 100)
mapped = df[df.reference_name != -1]
unmapped = df[df.reference_name == -1]
pylab.hist(mapped.reference_length,
bins=bins,
alpha=0.5,
label="mapped {}".format(len(mapped)),
density=False)
pylab.hist(unmapped.reference,
bins=bins,
alpha=0.5,
label="unmapped {}".format(len(unmapped)),
density=False)
pylab.xlabel("Isoform length")
pylab.legend()
def hist_transcript(self, hide_unmapped=True):
pylab.clf()
if hide_unmapped is True:
query = "reference_length>0 and reference_name!=-1"
else:
query = "reference_length>0"
print(query)
ts = self.df.query(query).groupby(
"reference_name").count().reference_length
if len(ts) == 0:
print("nothing to plot")
return ts
ts.plot(kind="bar", color="r")
try:
pylab.tight_layout()
except:
pass
return ts
def bar_mapq(self, logy=True, xmin=0, xmax=60, fontsize=12):
self.df.mapq.hist()
if logy:
pylab.semilogy()
pylab.xlim([xmin, xmax])
pylab.xlabel("Mapping quality", fontsize=fontsize)
def plot_sirv_by_group(self, title, shift=5, plot=False, mapq_min=-1):
aa = self.df.query("reference_name not in [-1, '-1']").copy()
if len(aa) == 0:
return pd.Series(), self.df
aa['group'] = aa.reference_name.apply(lambda x: x[0:shift])
mapped = aa.query("mapq>@mapq_min").groupby("group").count()["mapq"]
mapped.name = None
if plot:
mapped.plot(kind="bar")
pylab.title(title)
pylab.tight_layout()
#data.to_csv(path + "_hq_sirv_grouped.csv")
return mapped, self.df
"""
read length histograms pacbio data
=====================================
QC pacbio example
"""
########################################
# First, let us get a data set example.
# Note the .bam extension
from sequana import sequana_data
dataset = sequana_data("test_pacbio_subreads.bam")
#############################################
# Create a :class:`sequana.pacbio.BAMPacbio` instance
from sequana.pacbio import PacbioSubreads
qc = PacbioSubreads(dataset)
#########################################
# plot the histogram of read length
qc.hist_read_length()
#################################################
# plot the histogram of the SNRs for each base
qc.hist_snr()
def test_pacbio_stride():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
with TempFile() as fh:
b.stride(fh.name, stride=2)
with TempFile() as fh:
b.stride(fh.name, stride=2, random=True)
def test_pacbio():
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
assert len(b) == 130
b.df
#assert b.nb_pass[1] == 130
with TempFile() as fh:
b.filter_length(fh.name, threshold_min=500)
print(b) # check length
assert b.stats['mean_GC'] > 62.46
assert b.stats['mean_GC'] < 65.47
b.summary()
b = PacbioSubreads(sequana_data("test_pacbio_subreads.bam"))
# test hist_snr from scratch
b._df = None
b.hist_snr()
# test hist_len from scratch
b._df = None
b.hist_read_length()
b.hist_nb_passes()
b.get_mean_nb_passes()
# test from scratch
b._df = None
b.hist_GC()
# test from scratch
b._df = None
b.plot_GC_read_len()
# test from scratch
b._df = None
with TempFile() as fh:
b.to_fasta(fh.name, threads=1)
with TempFile() as fh:
b.to_fastq(fh.name, threads=1)
with TempFile() as fh:
b.save_summary(fh.name)