Added Fixing Faulty SWISS-ID

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "rascore"
-version = "1.0.4"
+version = "1.0.5"
 description = "A tool for analyzing RAS protein structures"
 authors = ["mitch-parker <[email protected]>"]
 license = "Apache License 2.0"

src/rascore/util/constants/gene.py

@@ -26,8 +26,7 @@
 uniprot_acc_lst = ["P01116", "P01116-2", "P01112", "P01111"]
 
 gene_class_dict = {
-    "GTPase HRas": hras_name,
-    "GTPase KRas": kras_name,
-    "Isoform 2B of GTPase KRas": kras_name,
-    "GTPase NRas": nras_name,
+    "RASK_HUMAN":kras_name,
+    "RASN_HUMAN":nras_name, 
+    "RASH_HUMAN":hras_name,
 }

src/rascore/util/pages/query_page.py

@@ -142,9 +142,14 @@ def query_page():
 
         mask_dict[col] = st.sidebar.multiselect(
             rename_col_dict[col],
-            lst_col(mask_df, col, unique=True),
+            [x.upper() for x in lst_col(mask_df, col, unique=True)],
         )
 
+        if col == pdb_id_col:
+            mask_dict[col] = [f"{x[:4].lower()}{x[4:5]}" for x in mask_dict[col]]
+        elif col == pdb_code_col:
+            mask_dict[col] = [x.lower() for x in mask_dict[col]]
+
         mask_df = mask_st_table(mask_df, mask_dict)
 
     if len(mask_df) == 0:

src/rascore/util/pipelines/build_rascore.py

@@ -123,7 +123,7 @@
     match_class_col,
     gene_class_col,
     prot_class_col,
-    prot_col,
+    swiss_id_col,
     bio_lig_col,
     nuc_class_col,
     rcsb_path_col,
@@ -297,7 +297,7 @@ def update_annot(pdbaa_fasta_path, out_path=None, past_df=None, num_cpu=1):
                 match_class = pharm_match
             df.at[index, match_class_col] = match_class
 
-        df[gene_class_col] = df[prot_col].map(gene_class_dict)
+        df[gene_class_col] = df[swiss_id_col].map(gene_class_dict)
         df[nuc_class_col] = df[bio_lig_col].map(nuc_class_dict).fillna(gtp_name)
 
         df = annot_prot(
@@ -684,6 +684,7 @@ def build_rascore(out_path=None, pdbaa_fasta_path=None, num_cpu=1):
         search_pdbaa(
             pdbaa_fasta_path=pdbaa_fasta_path,
             search_lst=swiss_id_lst,
+            fix_dict={"7u8hA":"RASK_HUMAN","7u8hB":"RASK_HUMAN","7u8hC":"RASK_HUMAN","7u8hD":"RASK_HUMAN"},
             entry_table_path=entry_table_path,
             min_length=25,
         )

src/rascore/util/scripts/search_pdbaa.py

@@ -50,6 +50,7 @@
 def search_pdbaa(
     pdbaa_fasta_path,
     search_lst,
+    fix_dict=None,
     entry_table_path=None,
     min_length=None,
     min_resolution=None,
@@ -83,6 +84,10 @@ def search_pdbaa(
     for col in col_lst:
         record_dict[col] = list()
 
+    fix_lst = list()
+    if fix_dict is not None:
+        fix_lst = list(fix_dict.keys())
+
     for record in tqdm(
         list(pdbaa_dict.keys()), desc="Searching pdbaa", position=0, leave=True
     ):
@@ -107,6 +112,13 @@ def search_pdbaa(
             if pdb_id in search_lst:
                 get_seq = True
 
+        if len(fix_lst) > 0:
+            if not get_seq:
+                if pdb_id in fix_lst:
+                    fix_lst.remove(pdb_id)
+                    swiss_id = fix_dict[pdb_id]
+                    get_seq = True
+
         if get_seq:
 
             seq = get_record_seq(record)