def main():
data = load_data()
if args.edges == 0:
ppiGraph = nx.Graph(data.humanppi, name="HumanPPI")
else:
ppiGraph = nx.Graph(data.humanppi[0:args.edges], name="HumanPPI")
draw_kwargs = dict(node_size=5, font_size=6,
with_labels=False, linewidths=0.1, width=0.2)
layout_kw = dict(iterations=args.iterations)
img_title = 'Nodes: %i Edges: %i Iterations: %i Runtime: %s'
if args.iterations is None:
# Number of iters will be set to 10, check the get_iterations method.
iters = get_iterations()
for iteration in iters:
start_time = time.time()
fig = plt.figure(figsize=(10, 10))
ax = fig.add_subplot(111)
nx.draw(ppiGraph, pos=nx.spring_layout(
ppiGraph, iterations=iteration), **draw_kwargs)
run_time = time.time() - start_time
runtime_str = str(round(run_time, 2)) + ' seconds'
print("Iterations: %i Runt-time: %s" % (iteration, runtime_str))
subs = (len(ppiGraph), args.edges, iteration, runtime_str)
ax.set_title(img_title % subs)
filepath = 'report/graphs' + '/' + \
zeros(iteration, padlength=4) + '.png'
plt.savefig(filepath, bbox_inches='tight')
print("Done! Saved at : " + filepath)
fig.clf()
plt.close()
else:
start_time = time.time()
fig = plt.figure(figsize=(10, 10))
ax = fig.add_subplot(111)
nx.draw(ppiGraph, pos=nx.spring_layout(
ppiGraph, **layout_kw), **draw_kwargs)
runtime = time.time() - start_time
runtime_str = str(round(runtime, 2)) + ' seconds'
subs = (len(ppiGraph), args.edges, args.iterations, runtime_str)
ax.set_title(img_title % subs)
filepath = 'report/graphs' + '/' + \
zeros(args.iterations, padlength=4) + '.png'
plt.savefig(filepath, bbox_inches='tight')
print("Done! Saved at " + filepath)
fig.clf()
plt.close()
def plot_graph(graph, protein, Tc, nodecolor, nodesymbol):
fig = plt.figure(figsize=(20,20))
ax = fig.add_subplot(111)
drawkwargs = {'font_size': 10,\
'linewidths': 1,\
'width': 2,\
'with_labels': True,\
'node_size': 700,\
'node_color':nodecolor,\
'node_shape':'o',\
'style':'solid',\
'alpha':1,\
'cmap': mpl.cm.jet}
pos=nx.spring_layout(graph, iterations=200)
nx.draw(graph, pos, **drawkwargs )
if protein in Tc:
string = "Protein: %i Cancer"
string_s = (protein)
title=string%string_s
else:
string = "Protein: %i Non-Cancer"
string_s = (protein)
title=string%string_s
ax.set_title(title)
filepath = 'images'+'/'+zeros(protein, padlength=4)+'.jpg'
plt.savefig(filepath, bbox_inches='tight')
print 'Written to: '+filepath
fig.clf()
plt.close()
del fig
def make_function_graphs(data):
fgraph = nx.Graph(data.functions, name='Functions')
drawkwargs = {'font_size': 10,\
'linewidths': 1,\
'width': 2,\
'with_labels': True,\
'node_size': 700,\
'node_color':'w',\
'node_shape':'o',\
'style':'solid',\
'alpha':1,\
'cmap': mpl.cm.jet}
filepath = 'images'+'/functions/'
for function in data.functions_only:
g = nx.Graph()
nbrs = fgraph.neighbors(function)
edges_to_add = [tuple([function, i]) for i in nbrs]
g.add_edges_from(edges_to_add)
fig = plt.figure(figsize=(20,20))
ax = fig.add_subplot(111)
pos=nx.spring_layout(g, iterations=20)
nx.draw(g, pos, **drawkwargs )
ax.set_title(function)
filepathname = filepath+zeros(function, padlength=4)+'.jpg'
plt.savefig(filepathname, bbox_inches='tight')
print 'Written to: '+filepathname
fig.clf()
plt.close()
del fig
def plot_graph(graph, protein, Tc, nodecolor, nodesymbol):
fig = plt.figure(figsize=(20, 20))
ax = fig.add_subplot(111)
drawkwargs = {'font_size': 10,\
'linewidths': 1,\
'width': 2,\
'with_labels': True,\
'node_size': 700,\
'node_color':nodecolor,\
'node_shape':'o',\
'style':'solid',\
'alpha':1,\
'cmap': mpl.cm.jet}
pos = nx.spring_layout(graph, iterations=200)
nx.draw(graph, pos, **drawkwargs)
if protein in Tc:
string = "Protein: %i Cancer"
string_s = (protein)
title = string % string_s
else:
string = "Protein: %i Non-Cancer"
string_s = (protein)
title = string % string_s
ax.set_title(title)
filepath = 'images' + '/' + zeros(protein, padlength=4) + '.jpg'
plt.savefig(filepath, bbox_inches='tight')
print 'Written to: ' + filepath
fig.clf()
plt.close()
del fig
def make_function_graphs(data):
fgraph = nx.Graph(data.functions, name='Functions')
drawkwargs = {'font_size': 10,\
'linewidths': 1,\
'width': 2,\
'with_labels': True,\
'node_size': 700,\
'node_color':'w',\
'node_shape':'o',\
'style':'solid',\
'alpha':1,\
'cmap': mpl.cm.jet}
filepath = 'images' + '/functions/'
for function in data.functions_only:
g = nx.Graph()
nbrs = fgraph.neighbors(function)
edges_to_add = [tuple([function, i]) for i in nbrs]
g.add_edges_from(edges_to_add)
fig = plt.figure(figsize=(20, 20))
ax = fig.add_subplot(111)
pos = nx.spring_layout(g, iterations=20)
nx.draw(g, pos, **drawkwargs)
ax.set_title(function)
filepathname = filepath + zeros(function, padlength=4) + '.jpg'
plt.savefig(filepathname, bbox_inches='tight')
print 'Written to: ' + filepathname
fig.clf()
plt.close()
del fig
def eye(N, dtype=np.float32):
"""
Construct a 2D matrix with ones on the diagonal and zeros elsewhere.
Constructs a matrix in device memory whose diagonal elements
are set to 1 and non-diagonal elements are set to 0.
Parameters
----------
N : int
Number of rows or columns in the output matrix.
Returns
-------
e_gpu : pycuda.gpuarray.GPUArray
Diagonal matrix of dimensions `[N, N]` with diagonal values
set to 1.
Examples
--------
>>> import pycuda.driver as drv
>>> import pycuda.gpuarray as gpuarray
>>> import pycuda.autoinit
>>> import numpy as np
>>> import linalg
>>> linalg.init()
>>> N = 5
>>> e_gpu = linalg.eye(N)
>>> np.all(e_gpu.get() == np.eye(N))
True
>>> e_gpu = linalg.eye(v_gpu, np.complex64)
>>> np.all(e_gpu.get() == np.eye(N, np.complex64))
True
"""
if dtype not in [np.float32, np.float64, np.complex64, np.complex128]:
raise ValueError("unrecognized type")
if N <= 0:
raise ValueError("N must be greater than 0")
use_double = int(dtype in [np.float64, np.complex128])
use_complex = int(dtype in [np.complex64, np.complex128])
# Initialize output matrix:
e_gpu = misc.zeros((N, N), dtype)
# Get block/grid sizes:
dev = misc.get_current_device()
block_dim, grid_dim = misc.select_block_grid_sizes(dev, e_gpu.shape)
# Set this to False when debugging to make sure the compiled kernel is
# not cached:
cache_dir = None
eye_mod = SourceModule(eye_template.substitute(use_double=use_double, use_complex=use_complex), cache_dir=cache_dir)
eye = eye_mod.get_function("eye")
eye(e_gpu, np.uint32(N), block=block_dim, grid=grid_dim)
return e_gpu
def diag(v_gpu):
"""
Construct a diagonal matrix.
Constructs a matrix in device memory whose diagonal elements
correspond to the elements in the specified array; all
non-diagonal elements are set to 0.
Parameters
----------
v_obj : pycuda.gpuarray.GPUArray
Input array of length `n`.
Returns
-------
d_gpu : pycuda.gpuarray.GPUArray
Diagonal matrix of dimensions `[n, n]`.
Examples
--------
>>> import pycuda.driver as drv
>>> import pycuda.gpuarray as gpuarray
>>> import pycuda.autoinit
>>> import numpy as np
>>> import linalg
>>> linalg.init()
>>> v = np.array([1, 2, 3, 4, 5, 6], np.float32)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
>>> v = np.array([1j, 2j, 3j, 4j, 5j, 6j], np.complex64)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
"""
if v_gpu.dtype not in [np.float32, np.float64, np.complex64,
np.complex128]:
raise ValueError('unrecognized type')
if len(v_gpu.shape) > 1:
raise ValueError('input array cannot be multidimensional')
use_double = int(v_gpu.dtype in [np.float64, np.complex128])
use_complex = int(v_gpu.dtype in [np.complex64, np.complex128])
# Initialize output matrix:
d_gpu = misc.zeros((v_gpu.size, v_gpu.size), v_gpu.dtype)
# Get block/grid sizes:
dev = misc.get_current_device()
block_dim, grid_dim = misc.select_block_grid_sizes(dev, d_gpu.shape)
# Set this to False when debugging to make sure the compiled kernel is
# not cached:
cache_dir=None
diag_mod = \
SourceModule(diag_template.substitute(use_double=use_double,
use_complex=use_complex),
cache_dir=cache_dir)
diag = diag_mod.get_function("diag")
diag(v_gpu, d_gpu, np.uint32(v_gpu.size),
block=block_dim,
grid=grid_dim)
return d_gpu
def diag(v_gpu):
"""
Construct a diagonal matrix if input array is one-dimensional,
or extracts diagonal entries of a two-dimensional array.
--- If input-array is one-dimensional:
Constructs a matrix in device memory whose diagonal elements
correspond to the elements in the specified array; all
non-diagonal elements are set to 0.
--- If input-array is two-dimensional:
Constructs an array in device memory whose elements
correspond to the elements along the main-diagonal of the specified
array.
Parameters
----------
v_obj : pycuda.gpuarray.GPUArray
Input array of shape `(n,m)`.
Returns
-------
d_gpu : pycuda.gpuarray.GPUArray
---If v_obj has shape `(n,1)`, output is
diagonal matrix of dimensions `[n, n]`.
---If v_obj has shape `(n,m)`, output is
array of length `min(n,m)`.
Examples
--------
>>> import pycuda.driver as drv
>>> import pycuda.gpuarray as gpuarray
>>> import pycuda.autoinit
>>> import numpy as np
>>> import linalg
>>> linalg.init()
>>> v = np.array([1, 2, 3, 4, 5, 6], np.float32)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
>>> v = np.array([1j, 2j, 3j, 4j, 5j, 6j], np.complex64)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
>>> v = np.array([[1., 2., 3.],[4., 5., 6.]], np.float64)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> d_gpu
array([ 1., 5.])
"""
if v_gpu.dtype not in [np.float32, np.float64, np.complex64,
np.complex128]:
raise ValueError('unrecognized type')
if (len(v_gpu.shape) > 1) and (len(v_gpu.shape) < 3):
# Since CUDA assumes that arrays are stored in column-major
# format, the input matrix is assumed to be transposed:
n, m = v_gpu.shape
square = (n == m)
# Allocate the output array
d_gpu = gpuarray.empty(min(m, n), v_gpu.dtype.type)
diag_kernel = el.ElementwiseKernel("double *x, double *y, int z", "y[i] = x[(z+1)*i]", "diakernel")
diag_kernel(v_gpu,d_gpu,max(m,n))
return d_gpu
elif len(v_gpu.shape) >= 3:
raise ValueError('input array cannot have greater than 2-dimensions')
use_double = int(v_gpu.dtype in [np.float64, np.complex128])
use_complex = int(v_gpu.dtype in [np.complex64, np.complex128])
# Initialize output matrix:
d_gpu = misc.zeros((v_gpu.size, v_gpu.size), v_gpu.dtype)
# Get block/grid sizes:
dev = misc.get_current_device()
block_dim, grid_dim = misc.select_block_grid_sizes(dev, d_gpu.shape)
# Set this to False when debugging to make sure the compiled kernel is
# not cached:
cache_dir=None
diag_mod = \
SourceModule(diag_template.substitute(use_double=use_double,
use_complex=use_complex),
cache_dir=cache_dir)
diag = diag_mod.get_function("diag")
diag(v_gpu, d_gpu, np.uint32(v_gpu.size),
block=block_dim,
grid=grid_dim)
return d_gpu
def eye(N, dtype=np.float32):
"""
Construct a 2D matrix with ones on the diagonal and zeros elsewhere.
Constructs a matrix in device memory whose diagonal elements
are set to 1 and non-diagonal elements are set to 0.
Parameters
----------
N : int
Number of rows or columns in the output matrix.
Returns
-------
e_gpu : pycuda.gpuarray.GPUArray
Diagonal matrix of dimensions `[N, N]` with diagonal values
set to 1.
Examples
--------
>>> import pycuda.driver as drv
>>> import pycuda.gpuarray as gpuarray
>>> import pycuda.autoinit
>>> import numpy as np
>>> import linalg
>>> linalg.init()
>>> N = 5
>>> e_gpu = linalg.eye(N)
>>> np.all(e_gpu.get() == np.eye(N))
True
>>> e_gpu = linalg.eye(v_gpu, np.complex64)
>>> np.all(e_gpu.get() == np.eye(N, np.complex64))
True
"""
if dtype not in [np.float32, np.float64, np.complex64, np.complex128]:
raise ValueError('unrecognized type')
if N <= 0:
raise ValueError('N must be greater than 0')
use_double = int(dtype in [np.float64, np.complex128])
use_complex = int(dtype in [np.complex64, np.complex128])
# Initialize output matrix:
e_gpu = misc.zeros((N, N), dtype)
# Get block/grid sizes:
dev = misc.get_current_device()
block_dim, grid_dim = misc.select_block_grid_sizes(dev, e_gpu.shape)
# Set this to False when debugging to make sure the compiled kernel is
# not cached:
cache_dir = None
eye_mod = \
SourceModule(eye_template.substitute(use_double=use_double,
use_complex=use_complex),
cache_dir=cache_dir)
eye = eye_mod.get_function("eye")
eye(e_gpu, np.uint32(N), block=block_dim, grid=grid_dim)
return e_gpu
def diag(v_gpu):
"""
Construct a diagonal matrix.
Constructs a matrix in device memory whose diagonal elements
correspond to the elements in the specified array; all
non-diagonal elements are set to 0.
Parameters
----------
v_obj : pycuda.gpuarray.GPUArray
Input array of length `n`.
Returns
-------
d_gpu : pycuda.gpuarray.GPUArray
Diagonal matrix of dimensions `[n, n]`.
Examples
--------
>>> import pycuda.driver as drv
>>> import pycuda.gpuarray as gpuarray
>>> import pycuda.autoinit
>>> import numpy as np
>>> import linalg
>>> linalg.init()
>>> v = np.array([1, 2, 3, 4, 5, 6], np.float32)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
>>> v = np.array([1j, 2j, 3j, 4j, 5j, 6j], np.complex64)
>>> v_gpu = gpuarray.to_gpu(v)
>>> d_gpu = linalg.diag(v_gpu)
>>> np.all(d_gpu.get() == np.diag(v))
True
"""
if v_gpu.dtype not in [
np.float32, np.float64, np.complex64, np.complex128
]:
raise ValueError('unrecognized type')
if len(v_gpu.shape) > 1:
raise ValueError('input array cannot be multidimensional')
use_double = int(v_gpu.dtype in [np.float64, np.complex128])
use_complex = int(v_gpu.dtype in [np.complex64, np.complex128])
# Initialize output matrix:
d_gpu = misc.zeros((v_gpu.size, v_gpu.size), v_gpu.dtype)
# Get block/grid sizes:
dev = misc.get_current_device()
block_dim, grid_dim = misc.select_block_grid_sizes(dev, d_gpu.shape)
# Set this to False when debugging to make sure the compiled kernel is
# not cached:
cache_dir = None
diag_mod = \
SourceModule(diag_template.substitute(use_double=use_double,
use_complex=use_complex),
cache_dir=cache_dir)
diag = diag_mod.get_function("diag")
diag(v_gpu, d_gpu, np.uint32(v_gpu.size), block=block_dim, grid=grid_dim)
return d_gpu
def main():
data = load_data()
if args.edges == 0:
ppiGraph = nx.Graph(data.humanppi, name="HumanPPI")
else:
ppiGraph = nx.Graph(data.humanppi[0:args.edges], name="HumanPPI")
draw_kwargs = dict(node_size=5,
font_size=6,
with_labels=False,
linewidths=0.1,
width=0.2)
layout_kw = dict(iterations=args.iterations)
img_title = 'Nodes: %i Edges: %i Iterations: %i Runtime: %s'
if args.iterations == None:
iters = get_iterations()
for iteration in iters:
starttime = time.time()
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(111)
nx.draw(ppiGraph, pos=nx.spring_layout(ppiGraph, iterations=iteration ), **draw_kwargs )
runtime = time.time() - starttime
runtime_str = str(round(runtime, 2))+' seconds'
print "Iterations: %i Run-time: %s "%(iteration, runtime_str)
subs = (len(ppiGraph), args.edges, iteration, runtime_str)
ax.set_title(img_title%subs)
filepath = 'graphs'+'/'+zeros(iteration, padlength=4)+'.png'
plt.savefig(filepath, bbox_inches='tight')
print 'Written to: '+filepath
fig.clf()
plt.close()
else:
starttime = time.time()
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(111)
nx.draw(ppiGraph, pos=nx.spring_layout(ppiGraph, **layout_kw ), **draw_kwargs )
runtime = time.time() - starttime
runtime_str = str(round(runtime, 2))+' seconds'
subs = (len(ppiGraph), args.edges, args.iterations, runtime_str)
ax.set_title(img_title%subs)
filepath = 'graphs'+'/'+zeros(args.iterations, padlength=4)+'.png'
plt.savefig(filepath, bbox_inches='tight')
print 'Written to: '+filepath
fig.clf()
plt.close()