def filter_network(g, degree_threshold=None, largest_connected_component=True):
print "V,E:", g.number_of_nodes(), g.number_of_edges()
degrees = g.degree(with_labels=True)
subgraph_nodes = []
for id, d in degrees.iteritems():
if degree_threshold is None or d <= degree_threshold:
subgraph_nodes.append(id)
g_filtered = g.subgraph(subgraph_nodes)
if largest_connected_component:
component_nodes = networkx.connected_components(g_filtered)[0]
g_filtered = g_filtered.subgraph(component_nodes )
print "V,E filtered:", g_filtered.number_of_nodes(), g_filtered.number_of_edges()
return g_filtered
def filter_network(g, degree_threshold=None, largest_connected_component=True):
print "V,E:", g.number_of_nodes(), g.number_of_edges()
degrees = g.degree(with_labels=True)
subgraph_nodes = []
for id, d in degrees.iteritems():
if degree_threshold is None or d <= degree_threshold:
subgraph_nodes.append(id)
g_filtered = g.subgraph(subgraph_nodes)
if largest_connected_component:
component_nodes = networkx.connected_components(g_filtered)[0]
g_filtered = g_filtered.subgraph(component_nodes)
print "V,E filtered:", g_filtered.number_of_nodes(
), g_filtered.number_of_edges()
return g_filtered
def get_connected_components(G, return_as_graph_list=True):
"""
Finds (strongly in the case of directed network) connected components of graph
returnAsGraphList: returns list of graph objects corresponding to connected components (from larger to smaller)
otherwise returns list of node list corresponding nodes in connected components
"""
result_list = []
if return_as_graph_list:
result_list = networkx.connected_component_subgraphs(G)
else:
result_list = networkx.connected_components(G)
return result_list
def get_connected_components(G, return_as_graph_list=True):
"""
Finds (strongly in the case of directed network) connected components of graph
returnAsGraphList: returns list of graph objects corresponding to connected components (from larger to smaller)
otherwise returns list of node list corresponding nodes in connected components
"""
result_list = []
if return_as_graph_list:
result_list = networkx.connected_component_subgraphs(G)
else:
result_list = networkx.connected_components(G)
return result_list
def analyze_network(g, out_file=None, seeds=None, calculate_radius=False):
if out_file is None:
from sys import stdout
out_method = stdout.write
else:
file = open(out_file, "a")
out_method = file.write
# print networkx.info(g) # currently buggy but fixed in the next version of networkx
out_method("(V,E): %d %d\n" % (g.number_of_nodes(), g.number_of_edges()))
out_method("V/E: %f\n" %
(int(g.number_of_nodes()) / float(g.number_of_edges())))
degrees = g.degree()
degrees.sort()
out_method("Average degree: %f\n" %
float(sum(degrees) / float(len(degrees))))
out_method("Most connected 20 nodes: %s\n" % degrees[-20:])
connected_components = networkx.connected_components(g)
out_method("Connected component sizes: %s\n" %
map(len, connected_components))
# Radius calculation is very time consuming
if calculate_radius:
out_method("Radius (of the largest connected component): %d\n" %
get_network_radius(g.subgraph(connected_components[0])))
#print get_network_degree_histogram(g)
if seeds is not None:
if len(seeds) <= 1:
out_method("Average linker degree: %f\n" % (0))
out_method("Average seed connecting shortest paths length: %f\n" %
(0))
else:
node_to_ld = get_node_linker_degrees(g, seeds)
vals = [node_to_ld[v] for v in seeds]
out_method("Average linker degree: %f\n" %
(float(sum(vals)) / len(vals)))
#sp_lengths = networkx.all_pairs_shortest_path_length(g)
sum_length = 0
count = 0
for i, u in enumerate(seeds):
sp_lengths = networkx.single_source_dijkstra_path_length(g, u)
for j, v in enumerate(seeds):
if j < i:
sum_length += sp_lengths[v] #[u][v]
count += 1
out_method("Average seed connecting shortest paths length: %f\n" %
(float(sum_length) / count))
if out_file is not None:
file.close()
return
def analyze_network(g, out_file = None, seeds = None, calculate_radius = False):
if out_file is None:
from sys import stdout
out_method = stdout.write
else:
file = open(out_file, "a")
out_method = file.write
# print networkx.info(g) # currently buggy but fixed in the next version of networkx
out_method("(V,E): %d %d\n" % (g.number_of_nodes(), g.number_of_edges()))
out_method("V/E: %f\n" % (int(g.number_of_nodes()) / float(g.number_of_edges())))
degrees = g.degree()
degrees.sort()
out_method("Average degree: %f\n" % float(sum(degrees)/float(len(degrees))))
out_method("Most connected 20 nodes: %s\n" % degrees[-20:])
connected_components = networkx.connected_components(g)
out_method("Connected component sizes: %s\n" % map(len, connected_components))
# Radius calculation is very time consuming
if calculate_radius:
out_method("Radius (of the largest connected component): %d\n" % get_network_radius(g.subgraph(connected_components[0])))
#print get_network_degree_histogram(g)
if seeds is not None:
if len(seeds) <= 1:
out_method("Average linker degree: %f\n" % (0))
out_method("Average seed connecting shortest paths length: %f\n" % (0))
else:
node_to_ld = get_node_linker_degrees(g, seeds)
vals = [ node_to_ld[v] for v in seeds ]
out_method("Average linker degree: %f\n" % (float(sum(vals))/len(vals)))
#sp_lengths = networkx.all_pairs_shortest_path_length(g)
sum_length = 0
count = 0
for i, u in enumerate(seeds):
sp_lengths = networkx.single_source_dijkstra_path_length(g, u)
for j, v in enumerate(seeds):
if j<i:
sum_length += sp_lengths[v] #[u][v]
count += 1
out_method("Average seed connecting shortest paths length: %f\n" % (float(sum_length)/count))
if out_file is not None:
file.close()
return
def create_R_analyze_network_script(g,
seeds=None,
out_path="./",
title="",
scale_by_log=False):
if scale_by_log:
f = open(out_path + "analyze_network_log_scaled.r", "w")
f.write(
"postscript(\"%sanalyze_network_log_scaled.eps\", width = 6, height = 6, horizontal = FALSE, onefile = FALSE, paper = \"special\", title = \"%s\")\n"
% (out_path, title))
else:
f = open(out_path + "analyze_network.r", "w")
f.write(
"postscript(\"%sanalyze_network.eps\", width = 6, height = 6, horizontal = FALSE, onefile = FALSE, paper = \"special\", title = \"%s\")\n"
% (out_path, title))
f.write("par(mfrow=c(3,2))\n")
node_to_values = get_node_degree_related_values(g, seeds)
f.write(
"f<-function(l) { \nr<-c()\nfor(i in 1:length(l)) {\n if(l[i] <= 1) {\n r[i]<-l[i]\n }\n else { \n r[i]<-(1+log10(l[i]))\n } \n}\nreturn(r)\n}\n"
)
# Degree distribution
#degrees = g.degree(with_labels = True)
#n_node = len(degrees)
#d_max = max(degrees.values())
d_max = max(node_to_values.values(), key=lambda x: x[0])[0]
non_seed_degree_counts = [0] * (d_max + 1)
seed_degree_counts = [0] * (d_max + 1)
#for v,d in degrees.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#if d_ != d: print v, d, d_ # networkx includes self edges in degree
if v in seeds:
seed_degree_counts[d] += 1
else:
non_seed_degree_counts[d] += 1
f.write("n<-c(%s)\n" % ",".join(map(str, seed_degree_counts)))
f.write("N<-c(%s)\n" % ",".join(map(str, non_seed_degree_counts)))
f.write("s<-(n+N)\n")
if scale_by_log:
f.write("A<-matrix(c(f(n), f(N+n)-f(n)), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(A, yaxt=\"n\", names.arg=c(0:(length(n)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
else:
f.write("A<-matrix(c(n, N), nrow=2, byrow=TRUE)\n")
##f.write("barplot(A, yaxp=c(0,max(N+n),1), xaxp=c(1, length(n), 1), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n")
f.write(
"barplot(A, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(n)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n"
)
#f.write("barplot(A, xaxt=\"n\", legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n")
#f.write("axis(1, seq(0,length(n)-1,round(length(n)/5)), labels=seq(0,length(n)-1,round(length(n)/5)))\n")
# Log-log plot of degree distribution
#f.write("par(fig=c(0.6, 1, 0.1, 0.5), new = T)\n") # Make this plot inner
f.write("p<-c()\n")
f.write("d<-c()\n")
f.write(
"for(i in 2:length(s)) { if(s[i] != 0) { p[i]<-s[i]; d[i]<-i; } }\n")
f.write("fit<-lm(log10(d) ~ log10(p))\n")
f.write(
"plot(0:(length(s)-1), s, log=\"xy\", yaxt=\"n\", xlab=\"Degree (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, las=2)\n")
f.write(
"#plot(d, p, yaxt=\"n\", xlab=\"log(degree)\", ylab=\"log(# of nodes)\")\n"
)
f.write("#axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
f.write("abline(fit, col=2, lty=2)\n")
f.write("legend(\"topright\", c(\"linear (ls) fit\"), col=c(2), lty=2)\n")
#f.write("par(fig=c(0, 1, 0, 1))\n")
# Linker degree distribution
#node_to_ld = get_node_linker_degrees(g, seeds)
#ld_max = max(node_to_ld.values())
ld_max = max(node_to_values.values(), key=lambda x: x[1])[1]
non_seed_ldegree_counts = [0] * (ld_max + 1)
seed_ldegree_counts = [0] * (ld_max + 1)
#for v,ld in node_to_ld.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#d, ld_, d2, ld2 = node_to_values[v]
#assert (ld_ == ld)
#if ld_ != ld: print v, ld, ld_
if v in seeds:
seed_ldegree_counts[ld] += 1
else:
non_seed_ldegree_counts[ld] += 1
f.write("ln<-c(%s)\n" % ",".join(map(str, seed_ldegree_counts)))
f.write("lN<-c(%s)\n" % ",".join(map(str, non_seed_ldegree_counts)))
if scale_by_log:
f.write("lA<-matrix(c(f(ln), f(lN+ln)-f(ln)), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(lA, yaxt=\"n\", names.arg=c(0:(length(ln)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker Degree\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("#lA<-matrix(c(f(lN+ln), f(ln)), nrow=2, byrow=TRUE)\n")
f.write(
"#barplot(lA, beside=TRUE, yaxt=\"n\", names.arg=c(0:(length(ln)-1)), legend.text=rev(c(\"seed\",\"all\")), col=rev(heat.colors(2)), xlab=\"Linker Degree\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
else:
f.write("lA<-matrix(c(ln, lN), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(lA, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(ln)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker Degree\", ylab=\"Number of nodes\")\n"
)
# Linker degree ratio
non_seed_ld_ratio_counts = [0] * 11
seed_ld_ratio_counts = [0] * 11
#for v,ld in node_to_ld.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#ratio = float(ld) / degrees[v]
if ld == 0:
ratio = 0
else:
ratio = float(ld) / d
ratio = int(ratio * 10)
if v in seeds:
seed_ld_ratio_counts[ratio] += 1
else:
non_seed_ld_ratio_counts[ratio] += 1
f.write("rn<-c(%s)\n" % ",".join(map(str, seed_ld_ratio_counts)))
f.write("rN<-c(%s)\n" % ",".join(map(str, non_seed_ld_ratio_counts)))
if scale_by_log:
f.write("rA<-matrix(c(f(rn), f(rN+rn)-f(rn)), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(rA, yaxt=\"n\", names.arg=c(0:(length(rn)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree / Degree\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
else:
f.write("rA<-matrix(c(rn, rN), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(rA, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(rn)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree / Degree\", ylab=\"Number of nodes\")\n"
)
# Linker degree ratio at level 2
non_seed_ld2_ratio_counts = [0] * 11
seed_ld2_ratio_counts = [0] * 11
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
if ld2 == 0:
ratio = 0
else:
ratio = float(ld2) / d2
ratio = int(ratio * 10)
if v in seeds:
seed_ld2_ratio_counts[ratio] += 1
else:
non_seed_ld2_ratio_counts[ratio] += 1
f.write("rn2<-c(%s)\n" % ",".join(map(str, seed_ld2_ratio_counts)))
f.write("rN2<-c(%s)\n" % ",".join(map(str, non_seed_ld2_ratio_counts)))
if scale_by_log:
f.write(
"rA2<-matrix(c(f(rn2), f(rN2+rn2)-f(rn2)), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(rA2, yaxt=\"n\", names.arg=c(0:(length(rn2)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree level2 / Degree level2\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
else:
f.write("rA2<-matrix(c(rn2, rN2), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(rA2, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(rn2)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree level2 / Degree level2\", ylab=\"Number of nodes\")\n"
)
connected_components = networkx.connected_components(g)
connected_component_sizes = map(len, connected_components)
max_size = max(connected_component_sizes)
connected_component_seed_counts = [0] * (max_size + 1)
connected_component_non_seed_counts = [0] * (max_size + 1)
for connected_component in connected_components:
for i in connected_component:
if i in seeds:
connected_component_seed_counts[len(connected_component)] += 1
else:
connected_component_non_seed_counts[len(
connected_component)] += 1
f.write("c<-c(%s)\n" % ",".join(map(str, connected_component_seed_counts)))
f.write("C<-c(%s)\n" %
",".join(map(str, connected_component_non_seed_counts)))
if scale_by_log:
f.write("cA<-matrix(c(f(c), f(C+c)-f(c)), nrow=2, byrow=TRUE)\n")
#f.write("barplot(cA, yaxt=\"n\", log=\"x\", xaxp=c(0,length(c),1), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n")
f.write(
"barplot(cA, yaxt=\"n\", xaxt=\"n\", legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n"
)
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" %
",".join(map(str, [10**i for i in xrange(5)])))
f.write(
"axis(1, seq(1,length(c),round(length(c)/5)), labels=seq(1,length(c),round(length(c)/5)))\n"
)
else:
f.write("cA<-matrix(c(c, C), nrow=2, byrow=TRUE)\n")
f.write(
"barplot(cA, ylim=c(0,round(max(s)/4)), names.arg=c(1:length(c)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size\", ylab=\"Number of nodes\")\n"
)
f.write("mtext(\'%s\', outer=TRUE, line=-1)\n" % title)
f.write("dev.off()\n")
f.close()
return
def create_R_analyze_network_script(g, seeds = None, out_path = "./", title = "", scale_by_log=False):
if scale_by_log:
f = open(out_path + "analyze_network_log_scaled.r", "w")
f.write("postscript(\"%sanalyze_network_log_scaled.eps\", width = 6, height = 6, horizontal = FALSE, onefile = FALSE, paper = \"special\", title = \"%s\")\n" % (out_path, title))
else:
f = open(out_path + "analyze_network.r", "w")
f.write("postscript(\"%sanalyze_network.eps\", width = 6, height = 6, horizontal = FALSE, onefile = FALSE, paper = \"special\", title = \"%s\")\n" % (out_path, title))
f.write("par(mfrow=c(3,2))\n")
node_to_values = get_node_degree_related_values(g, seeds)
f.write("f<-function(l) { \nr<-c()\nfor(i in 1:length(l)) {\n if(l[i] <= 1) {\n r[i]<-l[i]\n }\n else { \n r[i]<-(1+log10(l[i]))\n } \n}\nreturn(r)\n}\n")
# Degree distribution
#degrees = g.degree(with_labels = True)
#n_node = len(degrees)
#d_max = max(degrees.values())
d_max = max(node_to_values.values(), key=lambda x: x[0])[0]
non_seed_degree_counts = [ 0 ] * (d_max + 1)
seed_degree_counts = [ 0 ] * (d_max + 1)
#for v,d in degrees.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#if d_ != d: print v, d, d_ # networkx includes self edges in degree
if v in seeds:
seed_degree_counts[d] += 1
else:
non_seed_degree_counts[d] += 1
f.write("n<-c(%s)\n" % ",".join(map(str, seed_degree_counts)))
f.write("N<-c(%s)\n" % ",".join(map(str, non_seed_degree_counts)))
f.write("s<-(n+N)\n")
if scale_by_log:
f.write("A<-matrix(c(f(n), f(N+n)-f(n)), nrow=2, byrow=TRUE)\n")
f.write("barplot(A, yaxt=\"n\", names.arg=c(0:(length(n)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
else:
f.write("A<-matrix(c(n, N), nrow=2, byrow=TRUE)\n")
##f.write("barplot(A, yaxp=c(0,max(N+n),1), xaxp=c(1, length(n), 1), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n")
f.write("barplot(A, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(n)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n")
#f.write("barplot(A, xaxt=\"n\", legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Degree\", ylab=\"Number of nodes\")\n")
#f.write("axis(1, seq(0,length(n)-1,round(length(n)/5)), labels=seq(0,length(n)-1,round(length(n)/5)))\n")
# Log-log plot of degree distribution
#f.write("par(fig=c(0.6, 1, 0.1, 0.5), new = T)\n") # Make this plot inner
f.write("p<-c()\n")
f.write("d<-c()\n")
f.write("for(i in 2:length(s)) { if(s[i] != 0) { p[i]<-s[i]; d[i]<-i; } }\n")
f.write("fit<-lm(log10(d) ~ log10(p))\n")
f.write("plot(0:(length(s)-1), s, log=\"xy\", yaxt=\"n\", xlab=\"Degree (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, las=2)\n")
f.write("#plot(d, p, yaxt=\"n\", xlab=\"log(degree)\", ylab=\"log(# of nodes)\")\n")
f.write("#axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
f.write("abline(fit, col=2, lty=2)\n")
f.write("legend(\"topright\", c(\"linear (ls) fit\"), col=c(2), lty=2)\n")
#f.write("par(fig=c(0, 1, 0, 1))\n")
# Linker degree distribution
#node_to_ld = get_node_linker_degrees(g, seeds)
#ld_max = max(node_to_ld.values())
ld_max = max(node_to_values.values(), key=lambda x: x[1])[1]
non_seed_ldegree_counts = [ 0 ] * (ld_max + 1)
seed_ldegree_counts = [ 0 ] * (ld_max + 1)
#for v,ld in node_to_ld.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#d, ld_, d2, ld2 = node_to_values[v]
#assert (ld_ == ld)
#if ld_ != ld: print v, ld, ld_
if v in seeds:
seed_ldegree_counts[ld] += 1
else:
non_seed_ldegree_counts[ld] += 1
f.write("ln<-c(%s)\n" % ",".join(map(str, seed_ldegree_counts)))
f.write("lN<-c(%s)\n" % ",".join(map(str, non_seed_ldegree_counts)))
if scale_by_log:
f.write("lA<-matrix(c(f(ln), f(lN+ln)-f(ln)), nrow=2, byrow=TRUE)\n")
f.write("barplot(lA, yaxt=\"n\", names.arg=c(0:(length(ln)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker Degree\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("#lA<-matrix(c(f(lN+ln), f(ln)), nrow=2, byrow=TRUE)\n")
f.write("#barplot(lA, beside=TRUE, yaxt=\"n\", names.arg=c(0:(length(ln)-1)), legend.text=rev(c(\"seed\",\"all\")), col=rev(heat.colors(2)), xlab=\"Linker Degree\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
else:
f.write("lA<-matrix(c(ln, lN), nrow=2, byrow=TRUE)\n")
f.write("barplot(lA, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(ln)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker Degree\", ylab=\"Number of nodes\")\n")
# Linker degree ratio
non_seed_ld_ratio_counts = [ 0 ] * 11
seed_ld_ratio_counts = [ 0 ] * 11
#for v,ld in node_to_ld.iteritems():
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
#ratio = float(ld) / degrees[v]
if ld == 0:
ratio = 0
else:
ratio = float(ld) / d
ratio = int(ratio * 10)
if v in seeds:
seed_ld_ratio_counts[ratio] += 1
else:
non_seed_ld_ratio_counts[ratio] += 1
f.write("rn<-c(%s)\n" % ",".join(map(str, seed_ld_ratio_counts)))
f.write("rN<-c(%s)\n" % ",".join(map(str, non_seed_ld_ratio_counts)))
if scale_by_log:
f.write("rA<-matrix(c(f(rn), f(rN+rn)-f(rn)), nrow=2, byrow=TRUE)\n")
f.write("barplot(rA, yaxt=\"n\", names.arg=c(0:(length(rn)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree / Degree\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
else:
f.write("rA<-matrix(c(rn, rN), nrow=2, byrow=TRUE)\n")
f.write("barplot(rA, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(rn)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree / Degree\", ylab=\"Number of nodes\")\n")
# Linker degree ratio at level 2
non_seed_ld2_ratio_counts = [ 0 ] * 11
seed_ld2_ratio_counts = [ 0 ] * 11
for v, values in node_to_values.iteritems():
d, ld, d2, ld2 = values
if ld2 == 0:
ratio = 0
else:
ratio = float(ld2) / d2
ratio = int(ratio * 10)
if v in seeds:
seed_ld2_ratio_counts[ratio] += 1
else:
non_seed_ld2_ratio_counts[ratio] += 1
f.write("rn2<-c(%s)\n" % ",".join(map(str, seed_ld2_ratio_counts)))
f.write("rN2<-c(%s)\n" % ",".join(map(str, non_seed_ld2_ratio_counts)))
if scale_by_log:
f.write("rA2<-matrix(c(f(rn2), f(rN2+rn2)-f(rn2)), nrow=2, byrow=TRUE)\n")
f.write("barplot(rA2, yaxt=\"n\", names.arg=c(0:(length(rn2)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree level2 / Degree level2\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
else:
f.write("rA2<-matrix(c(rn2, rN2), nrow=2, byrow=TRUE)\n")
f.write("barplot(rA2, ylim=c(0,round(max(s)/4)), names.arg=c(0:(length(rn2)-1)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Linker degree level2 / Degree level2\", ylab=\"Number of nodes\")\n")
connected_components = networkx.connected_components(g)
connected_component_sizes = map(len, connected_components)
max_size = max(connected_component_sizes)
connected_component_seed_counts = [ 0 ] * (max_size + 1)
connected_component_non_seed_counts = [ 0 ] * (max_size + 1)
for connected_component in connected_components:
for i in connected_component:
if i in seeds:
connected_component_seed_counts[len(connected_component)] += 1
else:
connected_component_non_seed_counts[len(connected_component)] += 1
f.write("c<-c(%s)\n" % ",".join(map(str, connected_component_seed_counts)))
f.write("C<-c(%s)\n" % ",".join(map(str, connected_component_non_seed_counts)))
if scale_by_log:
f.write("cA<-matrix(c(f(c), f(C+c)-f(c)), nrow=2, byrow=TRUE)\n")
#f.write("barplot(cA, yaxt=\"n\", log=\"x\", xaxp=c(0,length(c),1), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("barplot(cA, yaxt=\"n\", xaxt=\"n\", legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size (Log scale)\", ylab=\"Number of nodes (Log scale)\")\n")
f.write("axis(2, 0:5, labels=c(0,%s), las=2)\n" % ",".join(map(str, [ 10**i for i in xrange(5) ])))
f.write("axis(1, seq(1,length(c),round(length(c)/5)), labels=seq(1,length(c),round(length(c)/5)))\n")
else:
f.write("cA<-matrix(c(c, C), nrow=2, byrow=TRUE)\n")
f.write("barplot(cA, ylim=c(0,round(max(s)/4)), names.arg=c(1:length(c)), legend.text=c(\"seed\",\"all\"), col=heat.colors(2), xlab=\"Connected component size\", ylab=\"Number of nodes\")\n")
f.write("mtext(\'%s\', outer=TRUE, line=-1)\n" % title)
f.write("dev.off()\n")
f.close()
return
