remove includeinpanel

Author: rxu17

genie/database_to_staging.py

@@ -1802,7 +1802,7 @@ def stagingToCbio(
     bedDf = extract.get_syntabledf(
         syn,
         "SELECT Chromosome,Start_Position,End_Position,Hugo_Symbol,ID,"
-        "SEQ_ASSAY_ID,Feature_Type,includeInPanel,clinicalReported FROM"
+        "SEQ_ASSAY_ID,Feature_Type,clinicalReported FROM"
         f" {bedSynId} where CENTER in ('{center_query_str}')",
     )
 

genie_registry/bed.py

@@ -456,9 +456,12 @@ def create_gene_panel(self, beddf, seq_assay_id, gene_panel_path, parentid):
         """
         LOGGER.info("CREATING GENE PANEL")
         if not beddf.empty:
-            exonsdf = beddf[beddf["Feature_Type"] == "exon"]
+            # exonsdf = beddf[beddf["Feature_Type"] == "exon"]
             # Only include genes that should be included in the panels
-            include_exonsdf = exonsdf[exonsdf["includeInPanel"]]
+            #include_exonsdf = exonsdf[exonsdf["includeInPanel"]]
+            
+            # Include all genes in the panels
+            include_exonsdf =  beddf[beddf["Feature_Type"] == "exon"]
             # Write gene panel
             null_genes = include_exonsdf["Hugo_Symbol"].isnull()
             unique_genes = set(include_exonsdf["Hugo_Symbol"][~null_genes])
@@ -631,11 +634,11 @@ def _validate(self, beddf):
             "Start_Position",
             "End_Position",
             "Hugo_Symbol",
-            "includeInPanel",
+            #"includeInPanel",
         ]
         if len(beddf.columns) < len(newcols):
             total_error += (
-                "BED file: Must at least have five columns in this "
+                "BED file: Must at least have four columns in this "
                 "order: {}. Make sure there are "
                 "no headers.\n".format(", ".join(newcols))
             )
@@ -675,15 +678,15 @@ def _validate(self, beddf):
                     "Hugo_Symbol column, not the strand column\n"
                 )
 
-            warn, error = process_functions.check_col_and_values(
-                beddf,
-                "includeInPanel",
-                [True, False],
-                filename="BED file",
-                required=True,
-            )
-            warning += warn
-            total_error += error
+            #warn, error = process_functions.check_col_and_values(
+            #    beddf,
+            #    "includeInPanel",
+            #    [True, False],
+            #    filename="BED file",
+            #    required=True,
+            #)
+            #warning += warn
+            #total_error += error
 
             if to_validate_symbol:
                 gene_position_table = self.syn.tableQuery("SELECT * FROM syn11806563")

templates/dashboardTemplate.Rmd

@@ -645,7 +645,7 @@ A user of the GENIE data pointed out that the expected gene count numbers in the
 
 ```{r genepanel_diff, echo=F}
 if (!is.null(assay_infodf) & !is.null(this_bed)) {
-  gene_panel_bed = this_bed[this_bed$Feature_Type == "exon" & this_bed$includeInPanel == "True" & this_bed$Hugo_Symbol != "", ]
+  gene_panel_bed = this_bed[this_bed$Feature_Type == "exon" & this_bed$Hugo_Symbol != "", ]
   # Get count of symbols per panel
   symbol_count_per_panel = table(gene_panel_bed$Hugo_Symbol, gene_panel_bed$SEQ_ASSAY_ID)
   # Get whether or not a panel contains a certain symbol

tests/test_bed.py

@@ -309,9 +309,9 @@ def test_missingcols_failure__validate(bed_class):
     emptydf = pd.DataFrame()
     error, warning = bed_class._validate(emptydf)
     expected_errors = (
-        "BED file: Must at least have five columns in this order: "
-        "Chromosome, Start_Position, End_Position, Hugo_Symbol, "
-        "includeInPanel. Make sure there are no headers.\n"
+        "BED file: Must at least have four columns in this order: "
+        "Chromosome, Start_Position, End_Position, Hugo_Symbol. "
+        "Make sure there are no headers.\n"
     )
     assert error == expected_errors
     assert warning == ""
@@ -353,8 +353,6 @@ def test_badinputs_failure__validate(bed_class):
         "Make sure there are no headers.\n"
         "BED file: The End_Position column must only be integers. "
         "Make sure there are no headers.\n"
-        "BED file: Please double check your includeInPanel column.  "
-        "This column must only be these values: True, False\n"
         "BED file: Please double check your Chromosome column.  "
         "This column must only be these values: {possible_vals}\n".format(
             possible_vals=", ".join(validate.ACCEPTED_CHROMOSOMES)