modification to use only good chia-pet projects

src/find_interaction_cluster/community_finder.py

@@ -122,7 +122,7 @@ def find_communities(graph: nx.Graph, project: str,
     logging.debug("Finding community ...")
     if not result_file.is_file():
         cmd = f"mpirun -np 1 {ConfigGraph.get_hipmcl_prog()} -M {outfile} " \
-              f"-I 1.2 -per-process-mem 32 -o {result_file}"
+              f"-I 1.2 -per-process-mem 50 -o {result_file}"
         sp.check_call(cmd, shell=True, stderr=sp.STDOUT)
     communities = get_communities(result_file)
     dic_community = {}
@@ -369,16 +369,7 @@ def get_projects(global_weight: int) -> List[str]:
     if global_weight != 0:
         return [f'Global-weight-{global_weight}']
     else:
-        cnx = sqlite3.connect(ConfigGraph.db_file)
-        c = cnx.cursor()
-        query = f"SELECT DISTINCT id_sample " \
-                f"FROM cin_projects " \
-                f"WHERE id_sample = 'GSM1872888' "
-        c.execute(query)
-        res = list(np.asarray(c.fetchall()).flatten())
-        c.close()
-        cnx.close()
-        return res
+        return ConfigGraph.good_projects
 
 
 def get_projects_name(global_weights: List[int]) -> Tuple[List[str], Dict]:

src/find_interaction_cluster/config.py

@@ -8,6 +8,7 @@ Description: Configuration folder
 
 from ..db_utils.config import Config
 from pathlib import Path
+from ..figures_utils.config_figures import get_good_project
 
 
 def get_weight_folder(weight: int, global_weight: int):
@@ -91,4 +92,5 @@ class ConfigGraph:
     hip_zip = data / 'hipMCL.zip'
     get_hip_folder = get_hipmcl_folder
     get_hipmcl_prog = get_hipmcl_prog
+    good_projects = get_good_project()
 

src/nt_composition/config.py

@@ -9,6 +9,7 @@ file that will be produced by this module.
 
 from ..db_utils.config import Config
 from typing import List
+from ..figures_utils.config_figures import get_good_project
 
 
 def get_weight_folder(weight: int, global_weight: int, ft_type: str):
@@ -149,3 +150,4 @@ class ConfigNt:
     selected_project = interaction / "selected_sample.txt"
     get_density_file = get_density_file
     get_features = get_features
+    good_projects = get_good_project()

src/nt_composition/make_nt_correlation.py

@@ -91,10 +91,12 @@ def get_project_colocalisation(cnx: sqlite3.Connection, project: str,
                         AND t1.exon2 = t3.id
                         AND t2.id_gene != t3.id_gene"""
     else:
+        good_projects = tuple(ConfigNt.good_projects)
         if same_gene:
             query = f"SELECT exon1, exon2{select_add} " \
                     f"FROM cin_exon_interaction " \
                     f"WHERE weight >= {weight} " \
+                    f"AND id_project IN {good_projects}" \
                     f"GROUP BY exon1, exon2 " \
                     f"HAVING COUNT(*) >= {global_weight}"
         else:
@@ -104,6 +106,7 @@ def get_project_colocalisation(cnx: sqlite3.Connection, project: str,
                         AND t1.exon1 = t2.id
                         AND t1.exon2 = t3.id
                         AND t2.id_gene != t3.id_gene
+                        AND t1.id_project IN {good_projects}
                         GROUP BY exon1, exon2
                         HAVING COUNT(*) >= {global_weight}"""
 
@@ -435,7 +438,7 @@ def create_all_frequency_figures(ps: int, weight: int = 1,
     logging_def(ConfigNt.interaction, __file__, logging_level)
     if global_weight == 0:
         di = pd.read_csv(ConfigNt.get_interaction_file(weight), sep="\t")
-        projects = di.loc[di['interaction_count'] > 300, 'projects'].values
+        projects = ConfigNt.good_projects
     else:
         di = pd.DataFrame()
         projects = [f"Global_projects_{global_weight}"]