Updated run_pipeline script to latest pipeline

deweylab · Mar 9, 2021 · f260f10 · f260f10
1 parent cf55653
commit f260f10
Showing 1 changed file with 53 additions and 38 deletions.
diff --git a/run_pipeline.py b/run_pipeline.py
@@ -18,10 +18,9 @@
 import map_sra_to_ontology
 from map_sra_to_ontology import ontology_graph
 from map_sra_to_ontology import load_ontology
-from map_sra_to_ontology import predict_sample_type
 from map_sra_to_ontology import config
-from map_sra_to_ontology import predict_sample_type
-from map_sra_to_ontology import run_sample_type_predictor
+import predict_sample_type
+from predict_sample_type import run_sample_type_predictor
 from predict_sample_type.learn_classifier import *
 from map_sra_to_ontology import pipeline_components as pc
 
@@ -44,7 +43,7 @@ def main():
         "EFO":"16",
         "CVCL":"4"}
     ont_id_to_og = {x:load_ontology.load(x)[0] for x in ont_name_to_ont_id.values()}
-    pipeline = p_48()
+    pipeline = p_53()
 
     all_mappings = []
     for tag_to_val in tag_to_vals:
@@ -114,36 +113,50 @@ def run_pipeline_on_key_vals(tag_to_val, ont_id_to_og, mapping_data):
 #    }
 #    return mappings
 
-def p_48():
-    spec_lex = pc.SpecialistLexicon(config.specialist_lex_location())
-    inflec_var = pc.SPECIALISTLexInflectionalVariants(spec_lex)
-    spell_var = pc.SPECIALISTSpellingVariants(spec_lex)
-    key_val_filt = pc.KeyValueFilter_Stage()
-    init_tokens_stage = pc.InitKeyValueTokens_Stage()
-    ngram = pc.NGram_Stage()
-    lower_stage = pc.Lowercase_Stage()
-    man_at_syn = pc.ManuallyAnnotatedSynonyms_Stage()
-    infer_cell_line = pc.InferCellLineTerms_Stage()
-    prop_spec_syn = pc.PropertySpecificSynonym_Stage()
-    infer_dev_stage = pc.ImpliedDevelopmentalStageFromAge_Stage()
-    linked_super = pc.LinkedTermsOfSuperterms_Stage()
-    cell_culture = pc.ConsequentCulturedCell_Stage()
-    filt_match_priority = pc.FilterOntologyMatchesByPriority_Stage()
-    real_val = pc.ExtractRealValue_Stage()
-    match_cust_targs = pc.ExactMatchCustomTargets_Stage()
-    cust_conseq = pc.CustomConsequentTerms_Stage()
-    delimit_plus = pc.Delimit_Stage('+')
-    delimit_underscore = pc.Delimit_Stage('_')
-    delimit_dash = pc.Delimit_Stage('-')
-    delimit_slash = pc.Delimit_Stage('/')
-    block_cell_line_key = pc.BlockCellLineNonCellLineKey_Stage()
-    subphrase_linked = pc.RemoveSubIntervalOfMatchedBlockAncestralLink_Stage()
-    cellline_to_implied_disease = pc.CellLineToImpliedDisease_Stage()
-    acr_to_expan = pc.AcronymToExpansion_Stage()
-    exact_match = pc.ExactStringMatching_Stage(["1", "2", "4", "5", "7", "8", "9"], query_len_thresh=3)
-    fuzzy_match = pc.FuzzyStringMatching_Stage(0.1, query_len_thresh=3)
-    two_char_match = pc.TwoCharMappings_Stage()
-    time_unit = pc.ParseTimeWithUnit_Stage()
+
+def p_53():
+    spec_lex = SpecialistLexicon(config.specialist_lex_location())
+    inflec_var = SPECIALISTLexInflectionalVariants(spec_lex)
+    spell_var = SPECIALISTSpellingVariants(spec_lex)
+    key_val_filt = KeyValueFilter_Stage()
+    init_tokens_stage = InitKeyValueTokens_Stage()
+    ngram = NGram_Stage()
+    lower_stage = Lowercase_Stage()
+    man_at_syn = ManuallyAnnotatedSynonyms_Stage()
+    infer_cell_line = InferCellLineTerms_Stage()
+    prop_spec_syn = PropertySpecificSynonym_Stage()
+    infer_dev_stage = ImpliedDevelopmentalStageFromAge_Stage()
+    linked_super = LinkedTermsOfSuperterms_Stage()
+    cell_culture = ConsequentCulturedCell_Stage()
+    filt_match_priority = FilterOntologyMatchesByPriority_Stage()
+    real_val = ExtractRealValue_Stage()
+    match_cust_targs = ExactMatchCustomTargets_Stage()
+    cust_conseq = CustomConsequentTerms_Stage()
+    delimit_plus = Delimit_Stage('+')
+    delimit_underscore = Delimit_Stage('_')
+    delimit_dash = Delimit_Stage('-')
+    delimit_slash = Delimit_Stage('/')
+    block_cell_line_key = BlockCellLineNonCellLineKey_Stage()
+    subphrase_linked = RemoveSubIntervalOfMatchedBlockAncestralLink_Stage()
+    cellline_to_implied_disease = CellLineToImpliedDisease_Stage()
+    acr_to_expan = AcronymToExpansion_Stage()
+    exact_match = ExactStringMatching_Stage(
+        [
+            "1",
+            "2",
+            "5",
+            "7",
+            "8",
+            "9",
+            "18" # Cellosaurus restricted to human cell lines
+        ],
+        query_len_thresh=3
+    )
+    fuzzy_match = FuzzyStringMatching_Stage(0.1, query_len_thresh=3)
+    two_char_match = TwoCharMappings_Stage()
+    time_unit = ParseTimeWithUnit_Stage()
+    prioritize_exact = PrioritizeExactMatchOverFuzzyMatch()
+    artifact_term_combo = TermArtifactCombinations_Stage()
 
     stages = [
         key_val_filt,
@@ -158,8 +171,8 @@ def p_48():
         spell_var,
         man_at_syn,
         acr_to_expan,
-        time_unit,
         exact_match,
+        time_unit,
         two_char_match,
         prop_spec_syn,
         fuzzy_match,
@@ -169,13 +182,15 @@ def p_48():
         cellline_to_implied_disease,
         subphrase_linked,
         cust_conseq,
+        artifact_term_combo,
         real_val,
         filt_match_priority,
         infer_cell_line,
         infer_dev_stage,
-        cell_culture]
-    return pc.Pipeline(stages, defaultdict(lambda: 1.0))
-
+        cell_culture,
+        prioritize_exact
+    ]
+    return Pipeline(stages, defaultdict(lambda: 1.0))
 
 if __name__ == "__main__":
     main()