def sortMatrix(hm, regionsFileName, transcriptID, transcript_id_designator):
"""
Iterate through the files noted by regionsFileName and sort hm accordingly
"""
labels = dict()
regions = []
defaultGroup = None
if len(regionsFileName) == 1:
defaultGroup = "genes"
for fname in regionsFileName:
fp = dti.openPossiblyCompressed(fname)
line = dti.getNext(fp)
labelColumn = None
while line.startswith("#"):
if not labelColumn:
labelColumn = dti.getLabel(line)
line = dti.getNext(fp)
# Find the label column
subtract = 0
if labelColumn is not None:
subtract = 1
# Determine the file type and load into a list (or list of lists)
cols = line.strip().split("\t")
if len(cols) - subtract < 3:
raise RuntimeError('{0} does not seem to be a recognized file type!'.format(fname))
elif len(cols) - subtract <= 6:
loadBED(line, fp, fname, labelColumn, labels, regions, defaultGroup)
elif len(cols) and dti.seemsLikeGTF(cols):
loadGTF(line, fp, fname, labels, regions, transcriptID, transcript_id_designator, defaultGroup)
else:
loadBED(line, fp, fname, labelColumn, labels, regions, defaultGroup)
fp.close()
# Do some sanity checking on the group labels and region names within them
s1 = set(hm.parameters['group_labels'])
for e in labels:
if e not in s1:
sys.exit("The computeMatrix output is missing the '{}' region group. It has [] but the specified regions have {}.\n".format(e, s1, labels.keys()))
# Make a dictionary out of current labels and regions
d = dict()
pos = 0
groupSizes = dict()
for idx, label in enumerate(hm.parameters['group_labels']):
s = hm.parameters['group_boundaries'][idx]
e = hm.parameters['group_boundaries'][idx + 1]
if label not in labels:
continue
d[label] = dict()
groupSize = 0
for reg in hm.matrix.regions[s:e]:
d[label][reg[2]] = pos
pos += 1
groupSize += 1
groupSizes[label] = groupSize
# Convert labels to an ordered list
labelsList = [""] * len(labels)
for k, v in labels.items():
labelsList[v] = k
# Reorder
order = []
boundaries = [0]
for idx, label in enumerate(labelsList):
# Make an ordered list out of the region names in this region group
_ = [""] * len(regions[idx])
for k, v in regions[idx].items():
_[v] = k
for name in _:
if name not in d[label]:
sys.stderr.write("Skipping {}, due to being absent in the computeMatrix output.\n".format(name))
continue
order.append(d[label][name])
boundaries.append(groupSizes[label] + boundaries[-1])
hm.matrix.regions = [hm.matrix.regions[i] for i in order]
order = np.array(order)
hm.matrix.matrix = hm.matrix.matrix[order, :]
# Update the parameters
hm.parameters["group_labels"] = labelsList
hm.matrix.group_labels = labelsList
hm.parameters["group_boundaries"] = boundaries
hm.matrix.group_boundaries = boundaries
def __init__(self, fnames, keepExons=False, labels=None, verbose=False):
"""
Driver function to actually parse files. The steps are as follows:
1) skip to the first non-comment line
2) Infer the type from that
3) Call a type-specific processing function accordingly
* These call the underlying C code for storage
* These handle chromsome name conversions (python-level)
Required inputs are as follows:
fnames: A list of (possibly compressed with gzip or bzip2) GTF or BED files.
Optional input is:
keepExons: For BED12 files, exons are ignored by default.
labels: Override the feature labels supplied in the file(s).
Note that this might instead be replaced later in the .features attribute.
verbose: Whether to print warnings (default: False)
"""
self.fname = []
self.filename = ""
self.chroms = []
self.features = []
self.tree = tree.initTree()
self.keepExons = keepExons
self.verbose = verbose
if not isinstance(fnames, list):
fnames = [fnames]
# Load the files
for labelIdx, fname in enumerate(fnames):
self.filename = fname
fp = openPossiblyCompressed(fname)
line, labelColumn = self.firstNonComment(fp)
if line is None:
# This will only ever happen if a file is empty or just has a header/comment
continue
line = line.strip()
ftype = self.inferType(fp, line, labelColumn)
if ftype != 'GTF' and labels is not None:
assert(len(labels) > labelIdx)
bname = labels[labelIdx]
else:
bname = basename(fname)
if ftype == 'GTF':
self.parseGTF(fp, line)
elif ftype == 'BED3':
self.parseBED(fp, line, 3, feature=bname, labelColumn=labelColumn)
elif ftype == 'BED6':
self.parseBED(fp, line, 6, feature=bname, labelColumn=labelColumn)
else:
self.parseBED(fp, line, 12, feature=bname, labelColumn=labelColumn)
fp.close()
# Sanity check
if self.tree.countEntries() == 0:
raise RuntimeError("None of the input BED/GTF files had valid regions")
if len(self.features) == 0:
raise RuntimeError("There were no valid feature labels!")
# vine -> tree
self.tree.finish()
def __init__(self, fnames, keepExons=False, labels=None, verbose=False):
"""
Driver function to actually parse files. The steps are as follows:
1) skip to the first non-comment line
2) Infer the type from that
3) Call a type-specific processing function accordingly
* These call the underlying C code for storage
* These handle chromsome name conversions (python-level)
Required inputs are as follows:
fnames: A list of (possibly compressed with gzip or bzip2) GTF or BED files.
Optional input is:
keepExons: For BED12 files, exons are ignored by default.
labels: Override the feature labels supplied in the file(s).
Note that this might instead be replaced later in the .features attribute.
verbose: Whether to print warnings (default: False)
"""
self.fname = []
self.filename = ""
self.chroms = []
self.features = []
self.tree = tree.initTree()
self.keepExons = keepExons
self.verbose = verbose
if not isinstance(fnames, list):
fnames = [fnames]
# Load the files
for labelIdx, fname in enumerate(fnames):
self.filename = fname
fp = openPossiblyCompressed(fname)
line, labelColumn = self.firstNonComment(fp)
if line is None:
# This will only ever happen if a file is empty or just has a header/comment
continue
line = line.strip()
ftype = self.inferType(fp, line, labelColumn)
if ftype != 'GTF' and labels is not None:
assert(len(labels) > labelIdx)
bname = labels[labelIdx]
else:
bname = basename(fname)
if ftype == 'GTF':
self.parseGTF(fp, line)
elif ftype == 'BED3':
self.parseBED(fp, line, 3, feature=bname, labelColumn=labelColumn)
elif ftype == 'BED6':
self.parseBED(fp, line, 6, feature=bname, labelColumn=labelColumn)
else:
self.parseBED(fp, line, 12, feature=bname, labelColumn=labelColumn)
fp.close()
# Sanity check
if self.tree.countEntries() == 0:
raise RuntimeError("None of the input BED/GTF files had valid regions")
if len(self.features) == 0:
raise RuntimeError("There were no valid feature labels!")
# vine -> tree
self.tree.finish()