fix: non-convervative merging + partial frag txs solution

modules/make_query_table.py

@@ -57,21 +57,20 @@ def query_table(path: str) -> pd.DataFrame:
     score["transcripts"] = [f"{k}.{v}" for k,v in zip(score["gene"], score["chain"])]
     score = score[["transcripts", "pred", "gene"]]
 
-    table = pd.merge(ortho_x_loss, score, left_on="transcript", right_on="transcripts", how="left")
+    table = pd.merge(ortho_x_loss, score, left_on="transcript", right_on="transcripts", how="outer")
+    table["helper"].fillna(table["gene"], inplace=True)
 
     # Create a new column with a rename orthology relationship
     table["relation"] = table["orthology_class"].map(Constants.ORTHOLOGY_TYPE)
     table["t_gene"].fillna(table["helper"].map(isoforms_dict), inplace=True)
+    table["transcripts"].fillna(table["transcript"], inplace=True)
 
     # Add paralog probabilities
     paralog = pd.merge(score, paralogs, on="transcripts").groupby("gene").agg({"pred": "max"}).reset_index()
     paralog.columns = ["helper", "paralog_prob"]
     table = pd.merge(table, paralog, on="helper", how="left")
     table["paralog_prob"].fillna(0, inplace=True)
 
-    # # Merge quality data
-    # table = pd.merge(table, quality, left_on="transcripts", right_on="Projection_ID")
-
     table = table[
         [
             "t_gene",
@@ -82,15 +81,13 @@ def query_table(path: str) -> pd.DataFrame:
             "pred",
             "q_gene",
             "paralog_prob",
-            # "confidence_level",
         ]
     ]
 
     info = [
         f"found {len(table)} projections, {len(table['helper'].unique())} unique transcripts, {len(table['t_gene'].unique())} unique genes",
         f"class stats: {table['class'].value_counts().to_dict()}",
         f"relation stats: {table['relation'].value_counts().to_dict()}"
-        # f"confidence stats: {table['confidence_level'].value_counts().to_dict()}",
     ]
 
     [log.record(i) for i in info]