def testFlyHiC():
fmcool1 = "tests/data/fly_hi-c/4DNFI8DRD739_bin100kb.cool"
fmcool2 = "tests/data/fly_hi-c/4DNFIZ1ZVXC8_bin100kb.cool"
cool1, binSize1 = readMcool(fmcool1, -1)
cool2, binSize2 = readMcool(fmcool2, -1)
#Check various .info() fields for consistency
assert coolerInfo(cool1, 'bin-size') == binSize1,\
f'coolerInfo() failed to retrieve metadata \'bin-size\' from {fmcool1}'
assert coolerInfo(cool2, 'bin-size') == binSize2,\
f'coolerInfo() failed to retrieve metadata \'bin-size\' from {fmcool2}'
assert coolerInfo(cool1, 'sum') == cool1.pixels()['count'][:].sum(),\
f'coolerInfo() failed to retrieve metadata \'sum\' from {fmcool1}'
assert coolerInfo(cool2, 'sum') == cool2.pixels()['count'][:].sum(),\
f'coolerInfo() failed to retrieve metadata \'sum\' from {fmcool2}'
assert coolerInfo(cool1, 'nbins') == cool1.bins().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nbins\' from {fmcool1}'
assert coolerInfo(cool2, 'nbins') == cool2.bins().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nbins\' from {fmcool2}'
assert coolerInfo(cool1, 'nnz') == cool1.pixels().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nnz\' from {fmcool1}'
assert coolerInfo(cool2, 'nnz') == cool2.pixels().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nnz\' from {fmcool2}'
assert coolerInfo(cool1, 'nchroms') == cool1.chroms().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nchroms\' from {fmcool1}'
assert coolerInfo(cool2, 'nchroms') == cool2.chroms().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nchroms\' from {fmcool2}'
def testHumanHiCInfo():
fmcool1 = "tests/data/human_hi-c/4DNFITKCX2DO.cool"
fmcool2 = "tests/data/human_hi-c/4DNFIQ5XCHDB.cool"
cool1, binSize1 = readMcool(fmcool1, -1)
cool2, binSize2 = readMcool(fmcool2, -1)
#Check various .info() fields for consistency
assert coolerInfo(cool1, 'bin-size') == binSize1,\
f'coolerInfo() failed to retrieve metadata \'bin-size\' from {fmcool1}'
assert coolerInfo(cool2, 'bin-size') == binSize2,\
f'coolerInfo() failed to retrieve metadata \'bin-size\' from {fmcool2}'
assert coolerInfo(cool1, 'sum') == cool1.pixels()['count'][:].sum(),\
f'coolerInfo() failed to retrieve metadata \'sum\' from {fmcool1}'
assert coolerInfo(cool2, 'sum') == cool2.pixels()['count'][:].sum(),\
f'coolerInfo() failed to retrieve metadata \'sum\' from {fmcool2}'
assert coolerInfo(cool1, 'nbins') == cool1.bins().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nbins\' from {fmcool1}'
assert coolerInfo(cool2, 'nbins') == cool2.bins().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nbins\' from {fmcool2}'
assert coolerInfo(cool1, 'nnz') == cool1.pixels().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nnz\' from {fmcool1}'
assert coolerInfo(cool2, 'nnz') == cool2.pixels().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nnz\' from {fmcool2}'
assert coolerInfo(cool1, 'nchroms') == cool1.chroms().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nchroms\' from {fmcool1}'
assert coolerInfo(cool2, 'nchroms') == cool2.chroms().shape[0],\
f'coolerInfo() failed to retrieve metadata \'nchroms\' from {fmcool2}'
def testHumanHiC():
fmcool1 = "tests/data/human_hi-c/4DNFITKCX2DO.cool"
fmcool2 = "tests/data/human_hi-c/4DNFIQ5XCHDB.cool"
binSize = 500000
h = 0
dBPMax = 5000000
bDownSample = False
cool1, _ = readMcool(fmcool1, -1)
cool2, _ = readMcool(fmcool2, -1)
#Test that the scc scores between a matrix and itself are always 1
results = hicrepSCC(cool1, cool1, h, dBPMax, bDownSample)
assert np.isclose(results, 1).all(),\
f"SCC scores between {fmcool1} and itself are not 1"
#Test that the scc scores agree with the previous R implementation
results = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample)
#Values given by R implementation of hicrep with same parameters
expected = np.array([
0.73050741, 0.67516601, 0.65743544, 0.74419469, 0.75864553, 0.75172288,
0.79556107, 0.66194007, 0.7141874, 0.78722237, 0.77622226, 0.77451858,
0.73061994, 0.70921468, 0.7447885, 0.75176337, 0.77104526, 0.83237602,
0.79166534, 0.80379132, 0.7504225, 0.64200014, 0.84293773, 0.79261671
])
assert np.isclose(results, expected).all(),\
f"SCC scores between {fmcool1} and {fmcool2} differ from those given by the R implementation"
def testHumanHiC():
fmcool1 = "tests/data/human_hi-c/4DNFITKCX2DO.cool"
fmcool2 = "tests/data/human_hi-c/4DNFIQ5XCHDB.cool"
binSize = 500000
h = 0
dBPMax = 5000000
bDownSample = False
cool1, _ = readMcool(fmcool1, -1)
cool2, _ = readMcool(fmcool2, -1)
#Test that the scc scores between a matrix and itself are always 1
results = hicrepSCC(cool1, cool1, h, dBPMax, bDownSample)
assert np.isclose(results, 1).all(),\
f"SCC scores between {fmcool1} and itself are not 1"
#Test that the scc scores agree with the previous R implementation
results = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample)
#Values given by R implementation of hicrep with same parameters
expected = np.array([
0.73050741, 0.67516601, 0.65743544, 0.74419469, 0.75864553, 0.75172288,
0.79556107, 0.66194007, 0.7141874, 0.78722237, 0.77622226, 0.77451858,
0.73061994, 0.70921468, 0.7447885, 0.75176337, 0.77104526, 0.83237602,
0.79166534, 0.80379132, 0.7504225, 0.64200014, 0.84293773, 0.79261671
])
assert np.isclose(results, expected).all(),\
f"SCC scores between {fmcool1} and {fmcool2} differ from those given by the R implementation"
# Test the computation of a subset of chromosomes give the same results as
# the whole set
chrNames = ['chr1', 'chr2', 'chrX']
resultsSub = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample, chrNames)
chrNamesAll = cool1.chroms()[:]['name'].tolist()
chrNamesAll = [name for name in chrNamesAll if name != 'M']
iChrs = np.where(np.isin(chrNamesAll, chrNames))[0]
assert (results[iChrs] == resultsSub).all(), f"""
SCC scores between {fmcool1} and {fmcool2} on chromosome subset
{chrNames} differ from those computed from the whole set. The whole
genome results are: {results} and the subset indices are {iChrs}.
"""
# Test the computation when excluding chromosomes give the same results as
# the whole set
exclNames = set(['chr1', 'chr2', 'chrX'])
resultsExcl = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample, chrNamesAll,
exclNames)
chrNamesRemain = [name for name in chrNamesAll if name not in exclNames]
ieChrs = np.where(np.isin(chrNamesAll, chrNamesRemain))[0]
assert (results[ieChrs] == resultsExcl).all(), f"""
def testFlyHiC():
fmcool1 = "tests/data/fly_hi-c/4DNFI8DRD739_bin100kb.cool"
fmcool2 = "tests/data/fly_hi-c/4DNFIZ1ZVXC8_bin100kb.cool"
binSize = -1
h = 1
dBPMax = 500000
bDownSample = False
cool1, binSize1 = readMcool(fmcool1, binSize)
cool2, binSize2 = readMcool(fmcool2, binSize)
assert coolerInfo(cool1, 'nbins') == coolerInfo(cool2, 'nbins'),\
f"Input cool files {fmcool1} and {fmcool2} have different number of bins"
assert binSize1 == binSize2,\
f"Input cool files {fmcool1} and {fmcool2} have different bin sizes"
assert coolerInfo(cool1, 'nchroms') == coolerInfo(cool2, 'nchroms'),\
f"Input cool files {fmcool1} and {fmcool2} have different number of chromosomes"
assert (cool1.chroms()[:] == cool2.chroms()[:]).all()[0],\
f"Input file {fmcool1} and {fmcool2} have different chromosome names"
results = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample)
expected = np.array([
9.936753824600870e-01, 9.950138992224218e-01, 9.951519844417879e-01,
9.935973973292749e-01, 9.933660605077106e-01, 9.927681695925705e-01,
6.238132870270471e-01
])
assert np.isclose(results, expected).all()
# Test the computation of a subset of chromosomes give the same results as
# the whole set
chrNames = ['chr2L', 'chr2R', 'chrX']
resultsSub = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample, chrNames)
chrNamesAll = cool1.chroms()[:]['name'].tolist()
chrNamesAll = [name for name in chrNamesAll if name != 'M']
iChrs = np.where(np.isin(chrNamesAll, chrNames))[0]
assert (results[iChrs] == resultsSub).all(), f"""
SCC scores between {fmcool1} and {fmcool2} on chromosome subset
{chrNames} differ from those computed from the whole set. The whole
genome results are: {results} and the subset indices are {iChrs}.
"""
# Test the computation when excluding chromosomes give the same results as
# the whole set
exclNames = set(['chr2L', 'chr2R', 'chrX'])
resultsExcl = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample, chrNamesAll,
exclNames)
chrNamesRemain = [name for name in chrNamesAll if name not in exclNames]
ieChrs = np.where(np.isin(chrNamesAll, chrNamesRemain))[0]
assert (results[ieChrs] == resultsExcl).all(), f"""
def testReadMcool():
fmcool = "tests/data/human_hi-c/4DNFICQK4N8B.mcool"
# Test that readMcool throws an exception if you try to read in an .mcool
# file at an unavailable resolution
with pytest.raises(KeyError):
readMcool(fmcool, 12345)
# Test that the read in .mcool file has the specified resolution and the
# correct chromosome labels
cool, binSize = readMcool(fmcool, 25000)
assert cool.binsize == 25000,\
f"Cooler object .mcool file {fmcool} has different resolution than requested"
fmcool = "tests/data/human_hi-c/4DNFIQ5XCHDB.cool"
# Test that readMcool throws an error if you try and read in a .cool file at
# a different resolution than it's builtin
with pytest.raises(KeyError):
readMcool(fmcool, 25000)
# Test that the read in .cool file has the correct built in binsize and the
# correct chromosome labels
cool, binSize = readMcool(fmcool, -1)
assert cool.binsize == 500000,\
f"Cooler object .mcool file {fmcool} has different resolution than requested"
def testFlyHiC():
fmcool1 = "tests/data/fly_hi-c/4DNFI8DRD739_bin100kb.cool"
fmcool2 = "tests/data/fly_hi-c/4DNFIZ1ZVXC8_bin100kb.cool"
binSize = -1
h = 1
dBPMax = 500000
bDownSample = False
cool1, binSize1 = readMcool(fmcool1, binSize)
cool2, binSize2 = readMcool(fmcool2, binSize)
assert coolerInfo(cool1, 'nbins') == coolerInfo(cool2, 'nbins'),\
f"Input cool files {fmcool1} and {fmcool2} have different number of bins"
assert binSize1 == binSize2,\
f"Input cool files {fmcool1} and {fmcool2} have different bin sizes"
assert coolerInfo(cool1, 'nchroms') == coolerInfo(cool2, 'nchroms'),\
f"Input cool files {fmcool1} and {fmcool2} have different number of chromosomes"
assert (cool1.chroms()[:] == cool2.chroms()[:]).all()[0],\
f"Input file {fmcool1} and {fmcool2} have different chromosome names"
results = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample)
expected = np.array([
9.936753824600870e-01, 9.950138992224218e-01, 9.951519844417879e-01,
9.935973973292749e-01, 9.933660605077106e-01, 9.927681695925705e-01,
6.238132870270471e-01
])
assert np.isclose(results, expected).all()
def main(*args):
import argparse
import subprocess
import re
np.random.seed(10)
parser = argparse.ArgumentParser()
parser.add_argument("fmcool1",
type=str,
help="First cooler multiple-binsize contact files")
parser.add_argument("fmcool2",
type=str,
help="Second cooler multiple-binsize contact files")
parser.add_argument(
"fout",
type=str,
help="Output results to this file. Output format would be\
one column of scc scores for each chromosome")
parser.add_argument(
"--binSize",
type=int,
default=-1,
help="Use this to select the bin size from the input mcool\
file. Default to -1, meaning that the inputs are treated as\
single-binsize .cool files")
parser.add_argument(
"--h",
type=int,
required=True,
help="Smooth the input contact matrices using a 2d mean\
filter with window size of 1 + 2 * value. This should\
be set according to the bin size. For example, you can try the\
following settings: --binSize=10000 --h=20,\
--binSize=25000 --h=10, --binSize=40000 --h5. Beware that\
these examples might not work in all cases and the user\
should adjust them according to the specific application"
)
parser.add_argument(
"--dBPMax",
type=int,
required=True,
help="Only consider contacts at most this number of bp away\
from the diagonal. For human genome, the value of\
5000000 was used in the original HiCRep paper.")
parser.add_argument(
"--bDownSample",
action='store_true',
default=False,
help="Down sample the input with more contact counts to\
the the same number of counts as the other input with less\
contact counts. If turned off, the input matrices will be\
normalized by dividing the counts by their respective total\
number of contacts.")
args = parser.parse_args()
header = "#" + " ".join(sys.argv) + "\n"
# Check if current script is under revision control
gitls = subprocess.Popen('cd ' +
os.path.dirname(os.path.realpath(__file__)) +
' && git ls-files --error-unmatch ' + __file__,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
encoding='utf-8')
if not gitls.stderr.read():
gitrev = subprocess.Popen('cd ' +
os.path.dirname(os.path.realpath(__file__)) +
' && git rev-parse HEAD --abbrev-ref HEAD',
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
encoding='utf-8')
if not gitrev.stderr.read():
p = re.compile("\n")
header += "# @rev " + p.sub("\n# @branch ",
gitrev.stdout.read().strip()) + "\n"
fmcool1 = args.fmcool1
fmcool2 = args.fmcool2
fout = args.fout
binSize = args.binSize
h = args.h
dBPMax = args.dBPMax
bDownSample = args.bDownSample
cool1, binSize1 = readMcool(fmcool1, binSize)
cool2, binSize2 = readMcool(fmcool2, binSize)
scc = hicrepSCC(cool1, cool2, h, dBPMax, bDownSample)
np.savetxt(fout, scc, "%30.15e", header=header)