Standardize TSV header casing; handle new DIRECT values; model_title …

…param
geneontology · Apr 9, 2020 · 6862ceb · 6862ceb
1 parent debc45e
commit 6862ceb
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 9 deletions.
diff --git a/pathway_connections.py b/pathway_connections.py
@@ -1,4 +1,5 @@
 import csv
+import itertools
 from ontobio.vocabulary.relations import OboRO
 from rdflib.term import URIRef
 from prefixcommons.curie_util import expand_uri
@@ -72,7 +73,7 @@ def __init__(self, id_a, id_b, mechanism, effect, direct, relation, pmid, linenu
             self.complex_b = COMPLEXES[self.id_b]
         # by default mechanism = molecular function
         mechanism_term = "GO:0003674"
-        if self.direct == "YES":
+        if self.direct in ["YES", "t"]:
             if(mechanism):
                 mechanism_term = MECHANISM_GO_MAPPING[mechanism]
         self.mechanism = { "name" : mechanism, "uri" : None, "term" : mechanism_term }
@@ -191,33 +192,39 @@ def full_statement_bnode_in_model(self, model):
                 if candidate_reg_triple in graph:
                     return candidate_reg_triple
 
+
+def upper_first(iterator):
+    return itertools.chain([next(iterator).upper()], iterator)
+
+
 class PathwayConnectionSet():
     def __init__(self, filename=None):
         self.connections = []
         linenum = 0
 
         if filename:
             with open(filename, "r") as f:
-                data = list(csv.DictReader(f, delimiter="\t"))
+
+                data = list(csv.DictReader(upper_first(f), delimiter="\t"))
                 for line in data:
                     linenum += 1
-                    print(line)
+                    # print(line)
 
                     # If up-regulates (including any variants of this), use RO:0002629 if DIRECT, and use RO:0002213 if not DIRECT or UNKNOWN
                     relation = None
                     if line["EFFECT"].startswith("up-regulates"):
-                        if line["DIRECT"] == "YES":
+                        if line["DIRECT"] in ["YES", "t"]:
                             relation = "RO:0002629"
-                        elif line["DIRECT"] == "NO":
+                        elif line["DIRECT"] in ["NO", "f"]:
                             relation = "RO:0002213"
                     # If down-regulates (including any variants of this), use RO:0002630 if DIRECT, and use RO:0002212 if not DIRECT or UNKNOWN
                     elif line["EFFECT"].startswith("down-regulates"):
-                        if line["DIRECT"] == "YES":
+                        if line["DIRECT"] in ["YES", "t"]:
                             relation = "RO:0002630"
-                        elif line["DIRECT"] == "NO":
+                        elif line["DIRECT"] in ["NO", "f"]:
                             relation = "RO:0002212"
                     # If unknown, use RO:0002211
-                    elif line["EFFECT"] == "unknown":
+                    elif line["EFFECT"] in ["unknown", ""]:
                         relation = "RO:0002211"
                     # If 'form complex', ignore these lines for now
                     elif line["EFFECT"] == "form complex":

diff --git a/pathway_importer.py b/pathway_importer.py
@@ -13,6 +13,8 @@
 parser = argparse.ArgumentParser()
 parser.add_argument('-f', "--filename", type=str, required=True,
                     help="Input filename of SIGNOR pathway data")
+parser.add_argument('-t', "--model_title",
+                    help="Model title. Defaults to --outfile value.")
 parser.add_argument('-o', "--outfile", type=str, required=True,
                     help="Output filename of generated model")
 
@@ -56,7 +58,11 @@ def main():
 
     args = parser.parse_args()
 
-    model = GoCamModel(args.outfile)
+    if args.model_title:
+        model_title = args.model_title
+    else:
+        model_title = args.outfile
+    model = GoCamModel(model_title)
     # p_connections = PathwayConnectionSet("SIGNOR-G2-M_trans_02_03_18.tsv")
     p_connections = PathwayConnectionSet(args.filename)
     linenum = 1

diff --git a/signor_complex.py b/signor_complex.py
@@ -89,6 +89,7 @@ def __init__(self, filename):
         self.families = self.grouping
 
 def main():
+    # TODO: parameterize SIGNOR_complexes.csv path
     complex_list = SignorComplexFactory("SIGNOR_complexes.csv")
     complex = complex_list.complexes["SIGNOR-C87"]
     complex_cc = "GO:0032991"
@@ -101,6 +102,7 @@ def main():
     pathway_line_mech_or_reg_activity = "pc.whatevs"
     # state "pathway_line_mech_or_reg_activity ENABLED_BY complex_cc"
 
+    # TODO: parameterize SIGNOR_PF.csv path
     pf_list = SignorProteinFamilyFactory("SIGNOR_PF.csv")
     pf = pf_list.families["SIGNOR-PF11"]
     print("PF list for " + pf.name + ":")