# get downsampling parameters noise, target = pcreode.get_thresholds( pca_reduced_data) # run pCreode out_graph, out_ids = pcreode.pCreode( data = pca_reduced_data, density = density, noise = noise, target = target, file_path = "/ti/workspace/.", num_runs = params["num_runs"], mute=True ) # score graphs, returns a vector of ranks by similarity graph ranks = pcreode.pCreode_Scoring(data = pca_reduced_data, file_path = "/ti/workspace/.", num_graphs = params["num_runs"], mute=True) # select most representative graph gid = graph_ranks[0] # extract cell graph # Wrapper's note: This is actually a cluster graph and a grouping, but none of the objects contain this grouping # the only thing that is available is a cell graph of only a subset of cells # so we use this cell graph as milestone network, and then project all cells onto this analysis = pcreode.Analysis( file_path = "/ti/workspace/.", graph_id = gid, data = pca_reduced_data, density = density, noise = params["noise"] )
dens = pcreode.Density(pca_reduced_data)
density_1 = dens.get_density(radius=1.0)
noise = 8.0
target = 50.0
file_path = './myeloid_w_ids/'
#pdb.set_trace()
out_graph, out_ids = pcreode.pCreode(data=pca_reduced_data,
density=density_1,
noise=noise,
target=target,
file_path=file_path,
num_runs=10)
pcreode.pCreode_Scoring(data=pca_reduced_data,
file_path=file_path,
num_graphs=10)
seed = 123
gid = 9
#Plot graph
pcreode.plot_save_graph(seed=seed,
file_path=file_path,
graph_id=gid,
data=pca_reduced_data,
overlay=data_raw.ELANE,
density=density_1,
file_out='Elane',
upper_range=1.5)
# get downsampling parameters noise, target = pcreode.get_thresholds( pca_reduced_data) # run pCreode out_graph, out_ids = pcreode.pCreode( data = pca_reduced_data, density = density, noise = noise, target = target, file_path = "/", num_runs = parameters["num_runs"], mute = True ) # score graphs, returns a vector of ranks by similarity graph_ranks = pcreode.pCreode_Scoring(data = pca_reduced_data, file_path = "/", num_graphs = parameters["num_runs"], mute=True) # select most representative graph gid = graph_ranks[0] # extract cell graph # Wrapper's note: This is actually a cluster graph and a grouping, but none of the objects contain this grouping # the only thing that is available is a cell graph of only a subset of cells # so we use this cell graph as milestone network, and then project all cells onto this analysis = pcreode.Analysis( file_path = "/", graph_id = gid, data = pca_reduced_data, density = density, noise = noise )
# downsample
downed, downed_ind = pcreode.Down_Sample(pca_reduced_data, density,
params["noise"], params["target"])
# run pCreode
out_graph, out_ids = pcreode.pCreode(data=pca_reduced_data,
density=density,
noise=params["noise"],
target=params["target"],
file_path="/.",
num_runs=params["num_runs"])
# score graphs
# Wrapper's note: There is currently no way of extracting the best graph ordering, even though it is printed. Will select random graph.
pcreode.pCreode_Scoring(data=pca_reduced_data,
file_path="/.",
num_graphs=params["num_runs"])
gid = np.random.choice(range(params["num_runs"]), 1)[0]
# extract cell graph
# Wrapper's note: This is actually a cluster graph and a grouping, but none of the objects contain this grouping
# the only thing that is available is a cell graph of only a subset of cells
# so we use this cell graph as milestone network, and then project all cells onto this
analysis = pcreode.Analysis(file_path="/.",
graph_id=gid,
data=pca_reduced_data,
density=density,
noise=params["noise"])
checkpoints["method_aftermethod"] = time.time()
file_path,
man_clust=[[0, 1], [50, 51]],
num_runs=num_runs,
mute=True)
# sparse cell types
out_graph, out_ids = pcreode.pCreode_sparse(pca_reduced_data,
density,
noise,
target,
file_path,
num_runs=num_runs,
mute=True)
# score graphs, returns a vector of ranks by similarity
graph_ranks = pcreode.pCreode_Scoring(data=pca_reduced_data,
file_path=file_path,
num_graphs=num_runs,
mute=True)
# select most representative graph
gid = graph_ranks[0]
# extract cell graph
analysis = pcreode.Analysis(file_path=file_path,
graph_id=gid,
data=pca_reduced_data,
density=density,
noise=noise)
#analysis.plot_save_graph( seed=0, overlay=expression.iloc[:,1], file_out=file_path, upper_range=3, node_label_size=0)
print "all good"