def haifeng_output_parser(self, row, argument=None, argument2=None):
"""
05-31-06 parse haifeng's output, only first two columns
based on fimbfs_parser()
2006-08-22 use ], [ as separator for edge_set
2006-08-29 cluster_no increments first(starting from 1)
2006-09-21 also parse the recurrence_array
"""
self.cluster_no += 1
self.cooccurrent_cluster_id += 1
cluster_list = []
gene_no2incidence_array = argument
curs = argument2
cluster = cluster_dstructure()
#initialize two sets
cluster.vertex_set = row[0][1:-1].split(',')
cluster.vertex_set = map(int, cluster.vertex_set)
cluster.vertex_set.sort()
cluster.cooccurrent_cluster_id = self.cooccurrent_cluster_id
cluster.cluster_id = self.cluster_no
cluster.edge_set = row[1][2:-2].split('], [')
for i in range(len(cluster.edge_set)):
cluster.edge_set[i] = cluster.edge_set[i].split(',')
cluster.edge_set[i] = map(int, cluster.edge_set[i])
cluster.edge_set[i].sort()
cluster.edge_set.sort()
cluster.no_of_edges = len(cluster.edge_set)
no_of_nodes = len(cluster.vertex_set)
cluster.splat_connectivity = 2*float(cluster.no_of_edges)/(no_of_nodes*(no_of_nodes-1))
cluster.connectivity = cluster.splat_connectivity
cluster.recurrence_array = row[2][1:-1].split(',')
cluster.recurrence_array = map(float, cluster.recurrence_array)
#05-31-06 fake the d_matrix
cluster.d_matrix = [[0,0,0]]
cluster.gim_array = get_vertex_set_gim_array(gene_no2incidence_array, cluster.vertex_set)
if self.debug:
print "cluster vertex_set: ", cluster.vertex_set
print "cluster edge_set: ", cluster.edge_set
print "cluster splat_connectivity: ", cluster.splat_connectivity
print "cluster recurrence_array: ", cluster.recurrence_array
print "cluster.d_matrix:", cluster.d_matrix
print "cluster.gim_array:", cluster.gim_array
raw_input("Continue?(Y/n)")
cluster_list.append(cluster)
return cluster_list
def fimbfs_parser(self, row, argument=None, argument2=None):
"""
10-28-05 similar to fim_parser(), but no sort for vertex_set and edge_set.
and has d_matrix. Output of MpiBFSCluster.py
12-06-05
add gene_no2incidence_array
calculate cluster.gim_array
"""
cluster_list = []
gene_no2incidence_array = argument #12-06-05
curs = argument2
cluster = cluster_dstructure()
#initialize two sets
cluster.vertex_set = row[0][1:-1].split(',')
cluster.vertex_set = map(int, cluster.vertex_set)
if len(cluster.vertex_set)<self.min_cluster_size: #pre-stop
return cluster_list
cluster.cooccurrent_cluster_id = self.cooccurrent_cluster_id
cluster.cluster_id = self.cluster_no
cluster.edge_set = row[1][2:-2].split('], [')
for i in range(len(cluster.edge_set)):
cluster.edge_set[i] = cluster.edge_set[i].split(',')
cluster.edge_set[i] = map(int, cluster.edge_set[i])
cluster.no_of_edges = len(cluster.edge_set)
no_of_nodes = len(cluster.vertex_set)
cluster.splat_connectivity = 2*float(cluster.no_of_edges)/(no_of_nodes*(no_of_nodes-1))
cluster.connectivity = cluster.splat_connectivity
cluster.recurrence_array = row[2][1:-1].split(',')
cluster.recurrence_array = map(float, cluster.recurrence_array)
cluster.d_matrix = row[3] #10-28-05 string form
cluster.gim_array = get_vertex_set_gim_array(gene_no2incidence_array, cluster.vertex_set) #12-06-05
if self.debug:
print "cluster vertex_set: ", cluster.vertex_set
print "cluster edge_set: ", cluster.edge_set
print "cluster splat_connectivity: ", cluster.splat_connectivity
print "cluster recurrence_array: ", cluster.recurrence_array
print "cluster.d_matrix:", cluster.d_matrix
print "cluster.gim_array:", cluster.gim_array #12-06-05
raw_input("Continue?(Y/n)")
cluster_list.append(cluster)
self.cluster_no += 1
self.cooccurrent_cluster_id += 1
return cluster_list