only one iteration of propagation, alpha fixed by PPI network edge nu…

…mber
0m1n0 · Dec 5, 2019 · 705b98a · 705b98a
1 parent 8737509
commit 705b98a
Show file tree

Hide file tree

Showing 2 changed files with 67 additions and 54 deletions.
diff --git a/stratipy/filtering_diffusion.py b/stratipy/filtering_diffusion.py
@@ -50,7 +50,7 @@ def propagation(M, adj, alpha, tol=10e-6):  # TODO equation, M, alpha
     adj : sparse matrix
         Adjacency matrice.
 
-    alpha : float, default: 0.7
+    alpha : float
         Diffusion/propagation factor with 0 <= alpha <= 1.
         For alpha = 0 : no diffusion.
         For alpha = 1 :
@@ -76,13 +76,16 @@ def propagation(M, adj, alpha, tol=10e-6):  # TODO equation, M, alpha
 
     X1 = M.astype(np.float32)
     X2 = alpha * X1.dot(A) + (1-alpha) * M
-    i = 0
-    while norm(X2-X1) > tol:
-        X1 = X2
-        X2 = alpha * X1.dot(A) + (1-alpha) * M
-        i += 1
-        print(' Propagation iteration = {}  ----- {}'.format(
-            i, datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")))
+
+    if tol:
+        i = 0
+        while norm(X2-X1) > tol:
+            X1 = X2
+            X2 = alpha * X1.dot(A) + (1-alpha) * M
+            i += 1
+            print(' Propagation iteration = {}  ----- {}'.format(
+                i, datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")),
+                  flush=True)
     return X2
 
 
@@ -151,9 +154,9 @@ def compare_ij_ji(ppi, out_min=True, out_max=True):
 
 
 # @profile
-def calcul_final_influence(M, adj, result_folder, influence_weight='min',
+def calcul_final_influence(M, adj, result_folder, alpha, influence_weight='min',
                            simplification=True, compute=False, overwrite=False,
-                           alpha=0.7, tol=10e-6):
+                           tol=10e-6):
     """Compute network influence score
 
     Network propagation iterative process is applied on PPI. (1) The  network
@@ -197,7 +200,7 @@ def calcul_final_influence(M, adj, result_folder, influence_weight='min',
         If True, new network influence score will be computed even if the file
         which same parameters already exists in the directory.
 
-    alpha : float, default: 0.7
+    alpha : float
         Diffusion (propagation) factor with 0 <= alpha <= 1.
         For alpha = 0 : no diffusion.
         For alpha = 1 :
@@ -224,7 +227,7 @@ def calcul_final_influence(M, adj, result_folder, influence_weight='min',
 
     existance_same_param = os.path.exists(final_influence_file)
     # TODO overwrite condition
-
+    print("+++++", final_influence_file)
     # check if same parameters file exists in directory
     if existance_same_param:
         final_influence_data = loadmat(final_influence_file)
@@ -239,6 +242,7 @@ def calcul_final_influence(M, adj, result_folder, influence_weight='min',
         if compute:
             # check if influence distance file exists
             existance_same_influence = os.path.exists(influence_distance_file)
+            print("+++++", influence_distance_file)
             if existance_same_influence:
                 influence_data = loadmat(influence_distance_file)
                 influence = influence_data['influence_distance']
@@ -496,7 +500,7 @@ def filtering(ppi_filt, result_folder, influence_weight, simplification,
               keep_singletons, min_mutation, max_mutation, mut_type):
     final_influence = (calcul_final_influence(
         sp.eye(ppi_filt.shape[0], dtype=np.float32), ppi_filt, result_folder,
-        influence_weight, simplification, compute, overwrite, alpha, tol))
+        alpha, influence_weight, simplification, compute, overwrite, tol))
 
     ppi_final, mut_final = filter_ppi_patients(
         result_folder, influence_weight, simplification, alpha, tol, ppi_total,

diff --git a/stratipy/parameters.py b/stratipy/parameters.py
@@ -7,45 +7,54 @@
 
 
 # TODO PPI type param
-param_grid = {'data_folder': ['../data/'],
-              'patient_data': ['SSC'],
-              'ssc_mutation_data': ['MAF50_LoF_mis15'],
-              # 'ssc_subgroups': ['SSC1_probands', 'SSC1_siblings',
-              #                   'SSC2_probands', 'SSC2_siblings'],
-              'ssc_subgroups': ['SSC1', 'SSC2'],
-              # 'ssc_subgroups': ['SSC1'],
-              # 'ssc_subgroups': ['SSC1', 'SSC2'],
-              # 'ssc_subgroups': ['SSC', 'SSC1', 'SSC2'],
-              # 'gene_data': ['all', 'pli', 'sfari', 'brain1SD', 'brain2SD'],
-              'gene_data': ['all', 'BrainSfariPli'],
-              'ppi_data': ['APID', 'STRING'],
-              # 'ppi_data': ['STRING'],
-              'influence_weight': ['min'],
-              'simplification': [True],
-              'compute': [True],
-              'overwrite': [False],
-              'alpha': [0.3],
-              'tol': [10e-3],
-              'ngh_max': [11],
-              'keep_singletons': [False],
-              'min_mutation': [0],
-              'max_mutation': [20000],
-              'mut_type': ['raw', 'diff', 'mean_qn', 'median_qn'],
-              # 'n_components': [2],
-              'n_components': range(2, 51),
-            #   'n_permutations': [1000],
-              'n_permutations': [300],
-              'sub_perm': [1],
-              'run_bootstrap': ['split'],
-              # 'run_bootstrap': ['full', 'split', None],
-              'run_consensus': [True],
-              'lambd': [0],
-              # 'lambd': [200],
-              # 'lambd': [0, 200],
-              'tol_nmf': [1e-3],
-              'compute_gene_clustering': [True],
-              'linkage_method': ['average'],
-              'p_val_threshold': [0.05]
+param_grid = {
+    'mut_type': ['raw', 'propagated', 'mean_qn', 'median_qn'],
+    # 'mut_type': ['raw'],
+
+    'ppi_data': ['APID', 'STRING'],
+    # 'ppi_data': ['STRING'],
+
+    # 'n_components': [2],
+     'n_components': range(2, 21),
+
+    # 'ssc_subgroups': ['SSC1_probands', 'SSC1_siblings', 'SSC2_probands', 'SSC2_siblings'],
+    # 'ssc_subgroups': ['SSC1', 'SSC2'],
+     'ssc_subgroups': ['SSC_all', 'SSC_male', 'SSC_female'],
+     # 'ssc_subgroups': ['SSC', 'SSC1', 'SSC2', 'SSC_all', 'SSC_male', 'SSC_female'],
+
+     # 'gene_data': ['all', 'pli', 'sfari', 'brain1SD', 'brain2SD'],
+     'gene_data': ['allGenes'],
+
+     'ssc_mutation_data': ['MAF1_LoF_mis15'],
+     # 'ssc_mutation_data': ['MAF1_LoF_mis30'],
+
+    'sub_perm': [1],
+    # 'run_bootstrap': ['split'],
+    'run_bootstrap': ['full'],
+
+    'lambd': [0],
+    # 'lambd': [200],
+    # 'lambd': [0, 200],
+
+    'data_folder': ['../data/'],
+    'patient_data': ['SSC'],
+    'influence_weight': ['min'],
+    'simplification': [True],
+    'compute': [True],
+    'overwrite': [False],
+    # 'alpha': [0.7],
+    'tol': [10e-3],
+    'ngh_max': [11],
+    'keep_singletons': [False],
+    'min_mutation': [0],
+    'max_mutation': [20000],
+    #   'n_permutations': [1000],
+    'n_permutations': [300],
+    'run_consensus': [True],
+    'tol_nmf': [1e-3],
+    'compute_gene_clustering': [True],
+    'linkage_method': ['average'],
+    'p_val_threshold': [0.05]
               }
 
 
@@ -65,7 +74,7 @@ def get_params(i):
     compute = d.get('compute')
     overwrite = d.get('overwrite')
 
-    alpha = d.get('alpha')
+    # alpha = d.get('alpha')
     tol = d.get('tol')
     ngh_max = d.get('ngh_max')
     keep_singletons = d.get('keep_singletons')
@@ -90,7 +99,7 @@ def get_params(i):
 
     return (data_folder, patient_data, ssc_mutation_data, ssc_subgroups,
             gene_data, ppi_data, influence_weight, simplification, compute,
-            overwrite, alpha, tol, ngh_max, keep_singletons, min_mutation,
+            overwrite, tol, ngh_max, keep_singletons, min_mutation,
             max_mutation, mut_type, n_components, n_permutations, sub_perm, sub_perm,
             run_bootstrap, run_consensus, lambd, tol_nmf,
             compute_gene_clustering, linkage_method, p_val_threshold)