2.1.0 updated

beaconplusWeb/package.json

@@ -20,6 +20,7 @@
     "@tippyjs/react": "^4.2.0",
     "animate": "^1.0.0",
     "animate.css": "^4.1.1",
+    "build": "^0.1.4",
     "classnames": "^2.2.6",
     "env-cmd": "^10.1.0",
     "file-saver": "^2.0.5",

bycon/config/variant_request_definitions.yaml

@@ -17,11 +17,6 @@ request_pars:
   - mate_end
   - variant_min_length
   - variant_max_length
-  # - gene_id
-  # - aminoacid_change
-  # - genomic_allele_short_form
-  # - cyto_bands
-  # - variant_query_digests
 
 # BeaconPlus
 multi_request_pars:

bycon/lib/beacon_response_generation.py

@@ -537,7 +537,7 @@ def __filtering_terms_query(self):
         if BYC["TEST_MODE"] is True:
             query = mongo_test_mode_query(self.ds_id, f_coll)
 
-        prdbug(f'filtering_terms_query: {query}')
+        # prdbug(f'filtering_terms_query: {query}')
 
         self.filtering_terms_query = query
 

bycon/lib/query_generation.py

@@ -3,7 +3,7 @@
 from os import environ
 from pymongo import MongoClient
 
-from bycon_helpers import days_from_iso8601duration, prdbug, return_paginated_list
+from bycon_helpers import days_from_iso8601duration, prdbug, prjsonnice, return_paginated_list
 from config import *
 from cytoband_parsing import Cytobands
 from genome_utils import ChroNames, GeneInfo, VariantTypes
@@ -189,23 +189,31 @@ def __preprocess_variant_pars(self):
         v_p_s = self.variant_request_definitions.get("request_pars", [])
         v_mp_s = self.variant_request_definitions.get("multi_request_pars", [])
 
+        vips = []
+
         # standard pars
+        s_q_p_0 = {}
         if (v_s_s := v_p_s & BYC_PARS.keys()):
-            s_q_p_0 = {}
             for v_p in v_s_s:
                 v_v = BYC_PARS.get(v_p)
                 s_q_p_0.update({ v_p: v_v })
                 BYC_VARGS.update({ v_p: v_v })
                 prdbug(f'...__preprocess_variant_pars: {v_p} {v_v}')
-            self.variant_multi_pars.append(s_q_p_0)
 
         for v_mp in v_mp_s:
             if len(v_mp_vs := BYC_PARS.get(v_mp, [])) > 0:
                 for v in v_mp_vs:
-                    self.variant_multi_pars.append({v_mp: v})
+                    if len(s_q_p_0.keys()) > 0:
+                        t_p = s_q_p_0.copy()
+                        t_p.update({v_mp: v})
+                        vips.append(self.__parse_variant_parameters(t_p))
+                    else:
+                        vips.append({v_mp: v})
+        if (len(vips)) < 1:
+            if len(s_q_p_0.keys()) > 0:
+                vips.append(self.__parse_variant_parameters(s_q_p_0))
 
-        for v_p_i, v_p_s in enumerate(self.variant_multi_pars):
-            self.variant_multi_pars[v_p_i] = self.__parse_variant_parameters(v_p_s)
+        self.variant_multi_pars = vips
 
 
     # -------------------------------------------------------------------------#
@@ -713,7 +721,7 @@ def __query_from_filters(self):
 
             # TODO: needs a general solution; so far for the iso age w/
             #       pre-calculated days field...
-            if "alphanumeric" in f_info.get("type", "ontology"):
+            if "alphanumeric" in f_info.get("ft_type", "ontology"):
                 f_class, comp, val = re.match(r'^(\w+):([<>=]+?)(\w[\w.]+?)$', f_info["id"]).group(1, 2, 3)
                 if "iso8601duration" in f_info.get("format", "___none___"):
                     val = days_from_iso8601duration(val)
@@ -737,7 +745,7 @@ def __query_from_filters(self):
                 if len(f_query_vals) == 1:
                     f_s_l.append({f_field: f_query_vals[0]})
                 else:
-                    if "alphanumeric" in f_infos[f_entity][f_field].get("type", "ontology"):
+                    if "alphanumeric" in f_infos[f_entity][f_field].get("ft_type", "ontology"):
                         q_l = []
                         for a_q_v in f_query_vals:
                             q_l.append({f_field: a_q_v})
@@ -775,20 +783,24 @@ def __query_from_filter_definitions(self, f_val):
         f_info = {
             "id": f_val,
             "scope": "biosamples",
-            "type": "___undefined___",
+            "ft_type": "___undefined___",
             "db_key": "___undefined___",
             "child_terms": [f_val]
         }
 
+        prdbug(f'...__query_from_filter_definitions: {f_val}')
+
         for f_d in f_d_s.values():
+            if f_d.get("collationed", False) is True:
+                continue
             f_re = re.compile(f_d.get("pattern", "___none___"))
             if f_re.match(f_val):
                 f_info = {
                     "id": f_val,
                     "scope": f_d.get("scope", "biosamples"),
                     "entity": f_d.get("entity", "biosample"),
                     "db_key": f_d["db_key"],
-                    "type": f_d.get("type", "ontology"),
+                    "ft_type": f_d.get("ft_type", "ontology"),
                     "format": f_d.get("format", "___none___"),
                     "child_terms": [f_val]
                 }

housekeepers/recordsSampler.py

@@ -46,7 +46,7 @@ def main():
         exit()
     set_entities()
 
-    example_queries = BYC["dataset_definitions"]["progenetix"].get("test_queries", {})
+    example_queries = BYC.get("test_queries", {})
 
     BYC_PARS.update({"response_entity_path_id":"analyses"})
     BYC_PARS.update({"inputfile":"___dummy___"})
@@ -87,7 +87,7 @@ def main():
             continue
 
         f_i_ids = ds["analyses.id"].get("target_values", [])
-        ana_ids.update(random_samples(f_i_ids, min(50, len(f_i_ids))))
+        ana_ids.update(random_samples(f_i_ids, min(BYC_PARS.get("limit", 200), len(f_i_ids))))
         print(f'==> now {len(ana_ids)}')
 
     BYC_PARS.update({"analysis_ids": list(ana_ids)})    

local/dataset_definitions.yaml

@@ -15,56 +15,6 @@ progenetix:
       - id: DUO:0000004
         label: no restriction
         version: 2021-11-11
-  # TODO: probably temporary ...
-  test_queries:
-    CDKN2AcnvQuery:
-      filters:
-        - NCIT:C3058
-      reference_name: refseq:NC_000009.12
-      start:
-        - 21000001
-        - 21975098
-      end:
-        - 21967753
-        - 23000000
-      variant_type: EFO:0030067
-    EIF4A1snvQuery:
-      reference_name: refseq:NC_000017.11
-      start:
-        - 7577120
-      alternate_bases: A
-      reference_bases: G
-    EIF4A1rangeQuery:
-      reference_name: refseq:NC_000017.11
-      start:
-        - 7572825
-      end:
-        - 7579005
-      variant_type: SO:0001059
-    # geneMatchQuery:
-    #   gene_id: CDK2
-    #   variant_min_length: 100000
-    #   variant_max_length: 200000
-    #   variant_type: EFO:0030070
-    HeLaIdentifierQuery:
-      filters:
-        - cellosaurus:CVCL_0030
-    CDKN2AshortFormExample:
-      filters:
-        - NCIT:C3058
-      variant_query_digests: 9:21000001-21975098--21967753-24000000:DEL
-    fusionExample:
-      variant_type: SO:0000806
-      reference_name: refseq:NC_000008.11
-      mate_name: refseq:NC_000012.12
-      start:
-        - 45100000
-      end:
-        - 47300000
-      mate_start:
-        - 26200000
-      mate_end:
-        - 35600000
 
 cellz:
   id: cellz

local/test_queries.yaml

@@ -1,18 +1,45 @@
-default:
-  CDKN2AcnvQuery:
-    filters: [NCIT:C3058]
-    reference_name: refseq:NC_000009.12
-    start: [21000001 21975098]
-    end: [21967753 23000000]
-    variant_type: EFO:0030067
-  EIF4A1snvQuery:
-    reference_name: refseq:NC_000017.11
-    start: [7577120]
-    variant_type: SO:0001059
-    alternate_bases: A
-    reference_bases: G
-  CDKN2ArangeQuery:
-    filters: [NCIT:C3058]
-    reference_name: refseq:NC_000009.12
-    start: [21967753]
-    end: [21975098]
+CDKN2AcnvQuery:
+  filters:
+    - NCIT:C3058
+  reference_name: refseq:NC_000009.12
+  start: [21000001,21975098]
+  end: [21967753,23000000]
+  variant_type: EFO:0030067
+EIF4A1snvQuery:
+  reference_name: refseq:NC_000017.11
+  start: [7577120]
+  variant_type: SO:0001059
+  alternate_bases: A
+  reference_bases: G
+EIF4A1rangeQuery:
+  reference_name: refseq:NC_000017.11
+  start: [7572825]
+  end: [7579005]
+  variant_type: SO:0001059
+geneMatchQuery:
+  gene_id: CDK2
+  variant_min_length: 100000
+  variant_max_length: 2000000
+  variant_type: EFO:0030070
+HeLaIdentifierQuery:
+  filters:
+    - cellosaurus:CVCL_0030
+CDKN2AshortFormExample:
+  filters:
+    - NCIT:C3058
+  variant_query_digests: 9:21000001-21975098--21967753-24000000:DEL
+fusionExample:
+  variant_type: SO:0000806
+  reference_name: refseq:NC_000008.11
+  mate_name: refseq:NC_000012.12
+  start: [45100000]
+  end: [47300000]
+  mate_start: [26200000]
+  mate_end: [35600000]
+ClinicalFiltersQuery:
+  filters:
+    - NCIT:C48786
+    - NCIT:C20197
+AgeAtDXquery:
+  filters:
+    - ageAtDiagnosis:<=P18Y