2.2.1

Major query stack rewrite; for more see https://bycon.progenetix.org/changes/
progenetix · Feb 21, 2025 · c6164cd · c6164cd
1 parent f6ef6dd
commit c6164cd
Show file tree

Hide file tree

Showing 108 changed files with 371,711 additions and 288,389 deletions.
diff --git a/MANIFEST.in b/MANIFEST.in
diff --git a/README.md b/README.md
diff --git a/beaconplusWeb/public/img/progenetix-logo.png b/beaconplusWeb/public/img/progenetix-logo.png
diff --git a/bycon/__init__.py b/bycon/__init__.py
@@ -1,22 +1,21 @@
 # __init__.py
 import sys, traceback
-from os import environ, path
-from pathlib import Path
+from os import path
 
 pkg_path = path.dirname( path.abspath(__file__) )
-bycon_lib_path = path.join( pkg_path, "lib" )
 sys.path.append( pkg_path )
-sys.path.append( bycon_lib_path )
 
 from config import *
 
+bycon_lib_path = path.join( pkg_path, "lib" )
+sys.path.append( bycon_lib_path )
+
+# try block to give at least some feedback on errors
 try:
 
     from beacon_auth import *
     from beacon_response_generation import *
     from bycon_helpers import *
-    from cytoband_parsing import *
-    from dataset_parsing import *
     from genome_utils import *
     from handover_generation import *
     from parameter_parsing import *
@@ -25,20 +24,26 @@
     from read_specs import *
     from response_remapping import *
     from schema_parsing import *
-    from service_utils import *
     from variant_mapping import *
 
-    # import byconServiceLibs
-
     read_service_definition_files()
     update_rootpars_from_local_or_HOST()
-    rest_path_elements()
-    set_beacon_defaults()
-    arguments_set_defaults()
-    parse_arguments()
-    set_entities()
-    initialize_bycon_service()
-    parse_filters()
+
+    if (defaults := BYC["beacon_defaults"].get("defaults", {})):
+        for d_k, d_v in defaults.items():
+            BYC.update( { d_k: d_v } )
+
+    ByconParameters().set_parameters()
+    ByconEntities().set_entities()
+    ByconDatasets().set_dataset_ids()
+
+    if (tm := BYC_PARS.get("test_mode")):
+        BYC.update({"TEST_MODE": test_truthy(tm)})
+    if BYC.get("TEST_MODE") is True:
+        BYC_PARS.update({"include_handovers": True})
+
+    set_user_name()
+    set_returned_granularities()    
 
 except Exception:
     if not "___shell___" in ENV:
@@ -49,3 +54,5 @@
     print(traceback.format_exc())
     print()
     exit()
+
+
diff --git a/bycon/byconServiceLibs/__init__.py b/bycon/byconServiceLibs/__init__.py
@@ -9,16 +9,16 @@
 sys.path.append( services_lib_path )
 
 try:
-    from bycon_bundler import *
+    from bycon_bundler import ByconBundler
+    from bycon_cluster import ByconCluster
+    from bycon_importer import ByconImporter
     from bycon_plot import *
-    from clustering_utils import *
     from collation_utils import *
     from cytoband_utils import *
     from datatable_utils import *
     from export_file_generation import *
     from file_utils import *
     from geoloc_utils import *
-    from importer_helpers import *
     from interval_utils import *
     from ontology_utils import *
     from service_helpers import *

diff --git a/bycon/byconServiceLibs/bycon_bundler.py b/bycon/byconServiceLibs/bycon_bundler.py
@@ -1,4 +1,6 @@
-import csv, datetime, re, sys
+import csv, re, sys
+
+from datetime import datetime
 
 from os import environ, path
 from pymongo import MongoClient
@@ -365,7 +367,7 @@ def __keyed_bundle_add_variants_from_lines(self):
             update_v = import_datatable_dict_line(update_v, fieldnames, v, "genomicVariant")
             update_v = ByconVariant().pgxVariant(update_v)
             update_v.update({
-                "updated": datetime.datetime.now().isoformat()
+                "updated": datetime.now().isoformat()
             })
             vars_ided[cs_id].append(update_v)
 
@@ -374,7 +376,7 @@ def __keyed_bundle_add_variants_from_lines(self):
             cs_ided[cs_id].update({"cnv_statusmaps": maps})
             cs_ided[cs_id].update({"cnv_stats": cs_cnv_stats})
             cs_ided[cs_id].update({"cnv_chro_stats": cs_chro_stats})
-            cs_ided[cs_id].update({"updated": datetime.datetime.now().isoformat()})
+            cs_ided[cs_id].update({"updated": datetime.now().isoformat()})
 
         self.keyedBundle.update({
             "individuals_by_id": inds_ided,

diff --git a/bycon/byconServiceLibs/bycon_cluster.py b/bycon/byconServiceLibs/bycon_cluster.py
@@ -0,0 +1,84 @@
+import scipy.cluster
+
+################################################################################
+################################################################################
+################################################################################
+
+class ByconCluster():
+    def __init__(self, plv):
+        self.plv = plv
+        self.data = plv.get("results", [])
+        self.cluster_metric = plv.get("plot_cluster_metric", "complete")
+        self.samples_cluster_type = plv.get("plot_samples_cluster_type", "")
+        self.tree_side = "right"
+        self.matrix = []
+        self.dendrogram = {}
+        self.new_order = []
+
+
+    # -------------------------------------------------------------------------#
+    # ----------------------------- public ------------------------------------#
+    # -------------------------------------------------------------------------#
+
+    def cluster_frequencies(self):
+        self.__matrix_from_interval_frequencies()
+        self.__make_dendrogram()
+        return self.dendrogram
+
+
+    # -------------------------------------------------------------------------#
+
+    def cluster_samples(self):
+        self.__matrix_from_samples()
+        self.__make_dendrogram()
+        return self.dendrogram
+
+
+    # -------------------------------------------------------------------------#
+    # ----------------------------- private -----------------------------------#
+    # -------------------------------------------------------------------------#
+
+    def __matrix_from_interval_frequencies(self):
+        for f_set in self.data:
+            i_f = f_set.get("interval_frequencies", [])
+            if_line = []
+            for i_f in f_set.get("interval_frequencies", []):
+                if_line.append( i_f.get("gain_frequency", 0) )
+            for i_f in f_set.get("interval_frequencies", []):
+                if_line.append( i_f.get("loss_frequency", 0) )
+            self.matrix.append(if_line)
+
+
+    # -------------------------------------------------------------------------#
+
+    def __matrix_from_samples(self):
+        for s in self.data:
+            s_line = []
+            if "intcoverage" in self.samples_cluster_type:
+                c_m = s.get("cnv_statusmaps", {})
+                dup_l = c_m.get("dup", [])
+                del_l = c_m.get("del", [])
+                for i_dup in dup_l:
+                    s_line.append(i_dup)
+                for i_del in del_l:
+                    s_line.append(i_del)
+            else:
+                c_s = s.get("cnv_chro_stats", {})
+                for c_a, c_s_v in c_s.items():
+                    s_line.append(c_s_v.get("dupfraction", 0))
+                for c_a, c_s_v in c_s.items():
+                    s_line.append(c_s_v.get("delfraction", 0))
+            self.matrix.append(s_line)
+
+
+    # -------------------------------------------------------------------------#
+
+    def __make_dendrogram(self):
+        linkage = scipy.cluster.hierarchy.linkage(self.matrix, method=self.cluster_metric)
+        self.new_order = scipy.cluster.hierarchy.leaves_list(linkage)
+        self.dendrogram = scipy.cluster.hierarchy.dendrogram(linkage, no_plot=True, orientation=self.tree_side)
+
+
+################################################################################
+################################################################################
+################################################################################