Dev cne

performance optimization and new version of manhattan plot (web interface)

FIX #68 FEAT #76 FEAT #77

jass/models/project.py

@@ -5,6 +5,7 @@ compute joint statistics and generate plots for a given set of phenotypes
 from __future__ import absolute_import
 from typing import List, Dict
 import os, sys
+import shutil
 import hashlib
 import traceback
 
@@ -285,14 +286,42 @@ def create_project_csv_file(worktable_path, csv_file, Nchunk):
 
 @app.task
 def create_project_data(
-    phenotype_ids, init_table_path, worktable_path, global_plot_path, quadrant_plot_path, csv_file=None
-):
-    try:
-        Nchunk = create_worktable_file(phenotype_ids, 
-            init_table_path, 
-            worktable_path, 
-            False, 
-            csv_file=csv_file)
+    phenotype_ids, 
+    init_table_path, 
+    worktable_path,
+    remove_nan = False,
+    stat = "jass.models.stats:omnibus_stat",
+    csv_file = None,
+    chunk_size = 50,
+    significance_treshold = 5*10**-8,
+    post_filtering = True,
+    delayed_gen_csv_file = False,
+    chromosome = None,
+    start = None,
+    end = None,
+    custom_loadings = None,
+    global_plot_path = None, 
+    quadrant_plot_path = None 
+    ):
+    
+    try:   
+        Nchunk = create_worktable_file(
+                phenotype_ids = phenotype_ids, 
+                init_file_path = init_table_path, 
+                project_hdf_path = worktable_path, 
+                remove_nan = remove_nan,
+                stat = stat,
+                optim_na = True,
+                csv_file = csv_file,
+                chunk_size = chunk_size,
+                significance_treshold = significance_treshold,
+                post_filtering = post_filtering,
+                delayed_gen_csv_file = delayed_gen_csv_file,
+                chromosome = chromosome,
+                pos_Start = start,
+                pos_End = end,
+                custom_loadings = custom_loadings
+                )
     except Exception as e:
         exc_type, exc_value, exc_traceback = sys.exc_info()
         log_path = get_file_building_tb_path(worktable_path)
@@ -300,9 +329,13 @@ def create_project_data(
         traceback.print_exception(exc_type, exc_value, exc_traceback, file=log_fh)
         log_fh.close()
         return
-    create_project_global_plot.delay(worktable_path, global_plot_path)
-    create_project_quadrant_plot.delay(worktable_path, quadrant_plot_path)
-    create_project_csv_file.delay(worktable_path, csv_file, Nchunk=Nchunk)
+    
+    if (global_plot_path is not None):
+        create_project_global_plot.delay(worktable_path, global_plot_path)
+    if (quadrant_plot_path is not None):
+        create_project_quadrant_plot.delay(worktable_path, quadrant_plot_path)
+    if (delayed_gen_csv_file and (csv_file is not None)):
+        create_project_csv_file.delay(worktable_path, csv_file, Nchunk=Nchunk)
 
 
 def create_project(phenotype_ids: List[str], available_phenotypes: List[Phenotype]):
@@ -321,11 +354,51 @@ def create_project(phenotype_ids: List[str], available_phenotypes: List[Phenotyp
     if project.status == Project.DOES_NOT_EXIST:
         os.makedirs(folder_path)
         create_project_data.delay(
-            phenotype_ids,
-            os.path.join(config["DATA_DIR"], "initTable.hdf5"),
-            project.get_worktable_path(),
-            project.get_global_manhattan_plot_path(),
-            project.get_quadrant_plot_path(),
-            csv_file=project.get_csv_path()
+            phenotype_ids=phenotype_ids,
+            init_table_path=os.path.join(config["DATA_DIR"], "initTable.hdf5"),
+            worktable_path=project.get_worktable_path(),
+            global_plot_path=project.get_global_manhattan_plot_path(),
+            quadrant_plot_path=project.get_quadrant_plot_path(),
+            csv_file=project.get_csv_path(),
+            delayed_gen_csv_file=True
+        )
+    return project
+ 
+ 
+def create_project_local(
+    phenotype_ids: List[str], 
+    available_phenotypes: List[Phenotype],
+    ip: str,
+    chromosome: str,
+    start: str = None,
+    end: str = None
+    ):
+    available_phenotype_ids = [phenotype.id for phenotype in available_phenotypes]
+    unavailable_requested_ids = set(phenotype_ids).difference(
+        set(available_phenotype_ids)
+    )
+    if len(unavailable_requested_ids) > 0:
+        raise Exception()  # FIXME with a nice exception
+    phenotypes = [
+        phenotype for phenotype in available_phenotypes if phenotype.id in phenotype_ids
+    ]
+    ip = ip.replace('.', '_')
+    id_project = "local_{}".format(ip)
+    project = Project(phenotypes=phenotypes, id=id_project)
+    folder_path = project.get_folder_path()
+    # If the folder exists, it is deleted with the files it contains
+    if os.path.exists(folder_path):
+        shutil.rmtree(folder_path)
+    # The folder is created
+    os.makedirs(folder_path)
+    create_project_data.delay(
+        phenotype_ids=phenotype_ids,
+        init_table_path=os.path.join(config["DATA_DIR"], "initTable.hdf5"),
+        worktable_path=project.get_worktable_path(),
+        csv_file=project.get_csv_path(),
+        chromosome=chromosome,
+        start=start,
+        end=end
         )
     return project
+    
\ No newline at end of file