Update code/pdb_featureVector.py

code/pdb_featureVector.py

@@ -230,35 +230,33 @@ def pdb(input_set, mode, impute):
                 modbase = no_swiss_models.copy()
                 print('Proceeding to  Modbase search...')
                 print('------------------------------------\n')
-                if len(modbase) > 0:
-                    modbase = modbase[SIMPLE_COLS]
-                    modbase.replace({'[]': np.NaN, 'nan-nan': np.NaN, '': np.NaN}, inplace=True)
-                    modbase = modbase.fillna(np.NaN)
-                    print('\n>> Adding Modbase residue positions...\n')
-                    modbase_simple = modbase[['uniprotID', 'wt', 'pos', 'mut','datapoint']]
-                    modbase_simple = modbase_simple.drop_duplicates(['uniprotID', 'wt', 'pos' ,'mut','datapoint'])
-                    modbaseOut, no_modbase_models_updated = addModbaseModels(modbase_simple, path_to_input_files, path_to_output_files)
+                
+                modbase = modbase[SIMPLE_COLS]
+                modbase.replace({'[]': np.NaN, 'nan-nan': np.NaN, '': np.NaN}, inplace=True)
+                modbase = modbase.fillna(np.NaN)
+                print('\n>> Adding Modbase residue positions...\n')
+                modbase_simple = modbase[['uniprotID', 'wt', 'pos', 'mut','datapoint']]
+                modbase_simple = modbase_simple.drop_duplicates(['uniprotID', 'wt', 'pos' ,'mut','datapoint'])
+                modbaseOut, no_modbase_models_updated = addModbaseModels(modbase_simple, path_to_input_files, path_to_output_files)
 
-                    if len(modbaseOut) > 0:
-                        modbase = modbase.merge(modbaseOut, on = ['uniprotID', 'wt', 'pos', 'mut','datapoint'], how = 'left')
-                        no_modbase_models_updated['sasa'] = np.NaN
-                        modbase.reset_index(inplace=True, drop=True)
-                        no_modbase_add = modbase[pd.isna(modbase.coordinates)]
-                        modbase = modbase[~pd.isna(modbase.coordinates)]
-                        no_modbase_models_updated = pd.concat([no_modbase_models_updated, no_modbase_add])
-                        print('\n>> Mapping to Modbase models...\n')
-                        modbase = changeUPtoModels(modbase)
-                        print('\n>> Calculating 3D distances for Modbase models...\n')
-                        modbase = isZeroDistance(modbase)
-                        modbase = match3DModels(modbase)
-                        modbase = selectMaxAnnot(modbase)
-                        modbase = modbase.sort_values(by=['datapoint', 'quality_score', 'distance','hitTotal', 'annotTotal'], ascending=[True, True, True, True, True])
-                        modbase = modbase.drop_duplicates(['datapoint'])
-                        modbase.replace({'[]': np.NaN, 'hit': 0.0}, inplace=True)
-                    else:
-                        modbase = pd.DataFrame(columns = SIMPLE_COLS)
+                if len(modbaseOut) > 0:
+                    modbase = modbase.merge(modbaseOut, on = ['uniprotID', 'wt', 'pos', 'mut','datapoint'], how = 'left')
+                    no_modbase_models_updated['sasa'] = np.NaN
+                    modbase.reset_index(inplace=True, drop=True)
+                    no_modbase_add = modbase[pd.isna(modbase.coordinates)]
+                    modbase = modbase[~pd.isna(modbase.coordinates)]
+                    no_modbase_models_updated = pd.concat([no_modbase_models_updated, no_modbase_add])
+                    print('\n>> Mapping to Modbase models...\n')
+                    modbase = changeUPtoModels(modbase)
+                    print('\n>> Calculating 3D distances for Modbase models...\n')
+                    modbase = isZeroDistance(modbase)
+                    modbase = match3DModels(modbase)
+                    modbase = selectMaxAnnot(modbase)
+                    modbase = modbase.sort_values(by=['datapoint', 'quality_score', 'distance','hitTotal', 'annotTotal'], ascending=[True, True, True, True, True])
+                    modbase = modbase.drop_duplicates(['datapoint'])
+                    modbase.replace({'[]': np.NaN, 'hit': 0.0}, inplace=True)
                 else:
-                    modbase = modbase[SIMPLE_COLS]
+                    modbase = pd.DataFrame(columns = SIMPLE_COLS)
     
             else:
                 no_modbase_models_updated = pd.DataFrame()
@@ -272,7 +270,15 @@ def pdb(input_set, mode, impute):
                     'lipidationBinary', 'siteBinary', 'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary', 'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
                     'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary', 'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary', 'modifiedResidueBinary', 'zincFingerBinary',
                     'motifBinary', 'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary', 'glycosylationBinary', 'propeptideBinary', 'sasa']
+            
+            if len(no_modbase_models_updated) == 0:
+                no_modbase_models_updated = pd.DataFrame()
+            no_modbase_models_updated = no_modbase_models_updated[~no_modbase_models_updated.datapoint.isin(modbase.datapoint.to_list())]
+            no_modbase_models_updated = no_modbase_models_updated[['uniprotID', 'wt', 'pos', 'mut', 'datapoint']]
+            no_modbase_models_updated.pos = no_modbase_models_updated.pos.astype(int)
             no_modbase_models_updated = no_modbase_models_updated.drop_duplicates()
+
+            
             if len(pdb)>0:
                 pdb = pdb[COLS]
                 pdb['Source'] = 'PDB'
@@ -288,8 +294,7 @@ def pdb(input_set, mode, impute):
                 modbase['Source'] = 'Modbase'
             else:
                 modbase = pd.DataFrame()
-            if len(no_modbase_models_updated) == 0:
-                no_modbase_models_updated = pd.DataFrame()
+            
     
             # st.write('======PDB==========')
             # st.write(pdb.to_string())
@@ -462,10 +467,15 @@ def pdb(input_set, mode, impute):
                 hours, rem = divmod(end - start, 3600)
                 minutes, seconds = divmod(rem, 60)
                 print("Time passed: {:0>2}:{:0>2}:{:05.2f}".format(int(hours), int(minutes), seconds))
+                if len(no_modbase_models_updated) >0 and (len(no_modbase_models_updated) !=org_len):
+                    st.write(f'{len(no_modbase_models_updated)} of {org_len} datapoins could not be mapped to any structures.')
+                    st.write(f'{org_len-len(no_modbase_models_updated)} of {org_len} datapoins were mapped to a structure.')
+                elif len(no_modbase_models_updated) == org_len:
+                    st.write(f'0 of {org_len} datapoins could not be mapped to any structures. Feature vector could not be created.')
 
                 return final_data
             elif len(no_modbase_models_updated) >0 and (len(no_modbase_models_updated) !=org_len):
                 st.write(f'{len(no_modbase_models_updated)} of {org_len} datapoins could not be mapped to any structures.')
-                st.write(f'{org_len-len(no_modbase_models_updated)} of {org_len} datapoins could not be mapped to any structures.')
+                st.write(f'{org_len-len(no_modbase_models_updated)} of {org_len} datapoins were mapped to a structure.')
             elif len(no_modbase_models_updated) == org_len:
-                st.write(f'0 of {org_len} datapoins could not be mapped to any structures. Feature vector could not be created.')
+                st.write(f'0 of {org_len} datapoins could not be mapped to any structures. Feature vector could not be created.')