#!/usr/bin/env python # coding: utf-8 import sys import time from simulation_functions import patients_data, nodes_position, generate_network, generate_network_shapes, addBetweenPathwaysConnection, save_dataset, plot_network_patient from simulation_nbs_functions import all_functions from confusion_matrix import minimal_element_0_to_1, simulated_confusion_matrix import datetime from sklearn.model_selection import ParameterGrid import networkx as nx import numpy as np import scipy.sparse as sp import pickle if (sys.version_info < (3, 2)): raise "Must be using Python ≥ 3.2" output_folder = "output/" # Pathways parameters pathwaysNum = 6 genesNum = 12 connProbability = 0.4 connNeighboors = 4 connBetweenPathways = 2 marker_shapes = ['o', 'p', '<', 'd', 's', '*'] # pathwaysNum = 10 # genesNum = 100 # connProbability = 0.4 # connNeighboors = 4 # connBetweenPathways = 2 # Simulate patients with a specific mutation profile patientsNum = 10 mutationProb = 0.2 print("\n------------ generate simulated data ------------ {}" .format(datetime.datetime.now() .strftime("%Y-%m-%d %H:%M:%S"))) PPI = generate_network_shapes( pathwaysNum, genesNum, connNeighboors, connProbability, marker_shapes) # PPI = generate_network( # pathwaysNum, genesNum, connNeighboors, connProbability) for BetweenPathwaysConnection in range(0, pathwaysNum*connBetweenPathways): PPI = addBetweenPathwaysConnection(PPI, pathwaysNum, genesNum) position = nodes_position(PPI, pathwaysNum, genesNum) patients, phenotypes = patients_data( patientsNum, PPI, genesNum, pathwaysNum, mutationProb) phenotype_idx = minimal_element_0_to_1(phenotypes) print(phenotype_idx) # save_dataset(PPI, position, patients, phenotypes, pathwaysNum, genesNum, # connProbability, connNeighboors, connBetweenPathways, patientsNum, # mutationProb, output_folder, new_data=True) ppi_data = 'simulation' influence_weight = 'min' simplification = True tol = 10e-3 keep_singletons = False compute = True overwrite = False min_mutation = 0 max_mutation = 100 n_components = 6 n_permutations = 1000 tol_nmf = 1e-3 linkage_method = 'average' phenotype_idx = phenotype_idx run_bootstrap = True run_consensus = True alpha = 0 ngh_max = 5 #10 qn = None #"mean", "median" lambd = 0 # 200 # all_functions(output_folder, ppi_data, patientsNum, mutationProb, # marker_shapes, pathwaysNum, genesNum, connNeighboors, # connProbability, connBetweenPathways, position, # influence_weight, simplification, # compute, overwrite, alpha, tol, ngh_max, keep_singletons, min_mutation, # max_mutation, qn, n_components, n_permutations, run_bootstrap, run_consensus, # lambd, tol_nmf, linkage_method) # print("\n------------ loading & formatting data ------------ {}" # .format(datetime.datetime.now() # .strftime("%Y-%m-%d %H:%M:%S"))) ppi_filt = nx.to_scipy_sparse_matrix(PPI, dtype=np.float32) ppi_final = ppi_filt mut_final = sp.csr_matrix(patients, dtype=np.float32) # simulated NBS parameters param_grid = {'output_folder': ['output/'], 'ppi_data': ['simulation'], 'ppi_filt': [ppi_filt], 'mut_final': [mut_final], 'influence_weight': ['min'], 'simplification': [True], 'compute': [True], 'overwrite': [False], 'alpha': [0, 0.5], 'tol': [10e-3], 'ngh_max': [5], #10 'keep_singletons': [False], 'min_mutation': [0], 'max_mutation': [100], 'qn': [None], 'n_components': [6], 'n_permutations': [1000], 'run_bootstrap': [True], 'run_consensus': [True], 'lambd': [0], # 200 'tol_nmf': [1e-3], 'linkage_method': ['average'] } start_all = time.time() for params in list(ParameterGrid(param_grid)): for i in params.keys(): exec("%s = %s" % (i, 'params[i]')) if alpha == 0 and qn is not None: print('############ PASS ############') pass else: print(qn) print(alpha) mut_type, mut_propag = all_functions(**params) plot_network_patient( mut_type, alpha, tol, PPI, position, np.array(mut_propag), patientsNum, phenotypes, marker_shapes, output_folder) # simulation_functions.plot_network_patient( # mut_type, alpha, tol, nx.from_scipy_sparse_matrix( # final_influence, create_using=nx.watts_strogatz_graph( # genesNum, connNeighboors, connProbability)), # position, patients, patientsNum, phenotypes, marker_shapes, # output_folder) end_all = time.time() print('---------- ALL = {} ---------- {}' .format(datetime.timedelta(seconds = end_all - start_all), datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))) # print("\n------------ confusion matrices ------------ {}" # .format(datetime.datetime.now() # .strftime("%Y-%m-%d %H:%M:%S"))) # # param_grid2 = {'output_folder': ['output/'], # 'influence_weight': ['min'], # 'simplification': [True], # 'alpha': [0], # 'tol': [10e-3], # 'ngh_max': [5], #[5, 10] # 'keep_singletons': [False], # 'min_mutation': [0], # 'max_mutation': [100], # 'qn': [None], # "mean", "median" # 'n_components': [6], # 'n_permutations': [1000], # 'lambd': [0], # 'tol_nmf': [1e-3], # 'linkage_method': ['average'], # 'phenotype_idx': [phenotype_idx], # 'pathwaysNum': [pathwaysNum] # } # for params in list(ParameterGrid(param_grid2)): # for i in params.keys(): # exec("%s = %s" % (i, 'params[i]')) # simulated_confusion_matrix(**params) # from scipy.io import loadmat # import networkx as nx # import matplotlib.pyplot as plt # from stratipy import filtering_diffusion # import scipy.sparse as sp # import numpy as np # # final_influence_directory = 'simulation/output/nbs/final_influence/' # final_influence_file = ( # final_influence_directory + # 'final_influence_PPI_simp=True_alpha=0.7_tol=0.01.mat') # final_influence_data = loadmat(final_influence_file) # final_influence = final_influence_data['final_influence_min'] # # final_influence.todense().shape # # a = nx.from_scipy_sparse_matrix(final_influence) # a # # mut_type, mut_propag = filtering_diffusion.propagation_profile( # sp.csr_matrix(patients, dtype=np.float32), # nx.to_scipy_sparse_matrix(PPI, dtype=np.float32), # result_folder = output_folder + 'nbs/', alpha=0.7, tol=10e-3, qn='mean') # mut_propag.shape # # #