change PDB/PDB-Dev ID case for files

ihm_validation/cx.py

@@ -68,7 +68,8 @@ def infinite_defaultdict(): return defaultdict(infinite_defaultdict)
 class CxValidation(GetInputInformation):
     def __init__(self, mmcif_file):
         super().__init__(mmcif_file)
-        self.ID = str(self.get_id())
+        self.ID = self.get_id()
+        self.ID_f = self.get_file_id()
         self.nos = self.get_number_of_models()
         self.dataset = self.get_dataset_comp()
         # Only atomic structures are supported so far
@@ -863,7 +864,7 @@ def plot_distograms_per_model_group(self, imgDirname='.'):
     def save_plots(self, plot, title, imgDirname='.'):
         imgpath = Path(
             imgDirname,
-            f'{self.ID}_{title}.html')
+            f'{self.ID_f}_{title}.html')
         save(
             plot, imgpath,
             resources=CDN,
@@ -872,7 +873,7 @@ def save_plots(self, plot, title, imgDirname='.'):
 
         imgpath_json = Path(
             imgDirname,
-            f'{self.ID}_{title}.json')
+            f'{self.ID_f}_{title}.json')
 
         with open(imgpath_json, 'w') as f:
             json.dump(json_item(plot, title), f)

ihm_validation/excludedvolume.py

@@ -21,7 +21,8 @@
 class GetExcludedVolume(GetInputInformation):
     def __init__(self, mmcif_file, cache):
         super().__init__(mmcif_file)
-        self.ID = Path(self.mmcif_file).stem
+        self.ID = self.get_id()
+        self.ID_f = self.get_file_id()
         self.nos = self.get_number_of_models()
         self.cache = cache
 
@@ -156,7 +157,7 @@ def get_exc_vol_given_sphere_parallel(self, sphere_list: list) -> (float, int):
     def run_exc_vol_parallel(self, model_dict: dict) -> dict:
         """ get exc vol info in parallel """
         # list_of_sphere_list=list(model_dict.values())
-        filename = str(Path(self.cache, self.ID + '_exv.txt'))
+        filename = str(Path(self.cache, self.ID_f + '_exv.txt'))
         if os.path.exists(filename):
             return self.process_exv(filename)
 

ihm_validation/get_plots.py

@@ -31,7 +31,8 @@
 class Plots(GetInputInformation):
     def __init__(self, mmcif, imageDirName, driver):
         super().__init__(mmcif)
-        self.ID = str(GetInputInformation.get_id(self))
+        self.ID = self.get_id()
+        self.ID_f = self.get_file_id()
         self.dirname = os.path.dirname(os.path.abspath(__file__))
         self.imageDirName = imageDirName
         self.filename = os.path.join(self.imageDirName)
@@ -41,7 +42,7 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
                                sas_data: dict, sas_fit: dict, cx_fit: dict) -> bokeh.plotting.figure:
 
         # create tabs list to add all the panel figures (model quality, data quality.. etc)
-        output_file(self.ID+"quality_at_glance.html", mode="inline")
+        output_file(self.ID_f+ "_" + "quality_at_glance.html", mode="inline")
 
         # MODEL QUALITY
         # check for molprobity or excluded volume data
@@ -99,7 +100,7 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
                 plots.append(p)
 
                 export_svg(p, filename=self.filename+'/' +
-                            self.ID+'_' + str(i) + "_quality_at_glance_MQ.svg", webdriver=self.driver)
+                            self.ID_f + "_" + str(i) + "_quality_at_glance_MQ.svg", webdriver=self.driver)
 
 
             grid = gridplot(plots, ncols=1,
@@ -166,7 +167,7 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
                 plots.append(p)
 
                 export_svg(p, filename=self.filename+'/' +
-                            self.ID+'_' + str(i) + "_quality_at_glance_MQ.svg", webdriver=self.driver)
+                            self.ID_f + "_" + str(i) + "_quality_at_glance_MQ.svg", webdriver=self.driver)
 
             grid = gridplot(plots, ncols=1,
                             merge_tools=True,
@@ -207,9 +208,9 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
         # make panel figures
         # first panel is model quality
         export_svg(fullplot, filename=self.filename+'/' +
-                    self.ID+"quality_at_glance_MQ.svg", webdriver=self.driver)
+                    self.ID_f + "_" +"quality_at_glance_MQ.svg", webdriver=self.driver)
         save(fullplot, filename=self.filename+'/' +
-             self.ID+"quality_at_glance_MQ.html")
+             self.ID_f + "_" +"quality_at_glance_MQ.html")
 
         # DATA QUALITY
         # check for sas data, if exists, plot
@@ -245,9 +246,9 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
             pd.title.align = "center"
             pd.output_backend = "svg"
             export_svg(pd, filename=self.filename+'/' +
-                        self.ID+"quality_at_glance_DQ.svg", webdriver=self.driver)
+                        self.ID_f + "_" +"quality_at_glance_DQ.svg", webdriver=self.driver)
             save(pd, filename=self.filename+'/' +
-                 self.ID+"quality_at_glance_DQ.html")
+                 self.ID_f + "_" +"quality_at_glance_DQ.html")
 
         # FIT TO DATA QUALITY
         # check for sas data, if exists, plot
@@ -285,6 +286,6 @@ def plot_quality_at_glance(self, molprobity_data: dict, exv_data: dict,
             pf.title.align = "center"
             pf.output_backend = "svg"
             export_svg(pf, filename=self.filename+'/' +
-                        self.ID+"quality_at_glance_FQ.svg", webdriver=self.driver)
+                        self.ID_f + '_' +"quality_at_glance_FQ.svg", webdriver=self.driver)
             save(pf, filename=self.filename+'/' +
-                 self.ID+"quality_at_glance_FQ.html")
+                 self.ID_f + '_' +"quality_at_glance_FQ.html")

ihm_validation/mmcif_io.py

@@ -91,7 +91,34 @@ def get_databases(self):
 
     def get_id(self):
         """Return _entry.id; Requires compliant CIF file"""
-        entry_id = self.system.id
+        ids = self.get_ranked_id_list()
+
+        if len(ids) == 0:
+            raise(ValueError('Missing system ID'))
+
+        id_type, entry_id = ids[0]
+
+        return entry_id
+
+    def get_file_id(self):
+        """Return _entry.id; Requires compliant CIF file"""
+        ids = self.get_ranked_id_list()
+
+        if len(ids) == 0:
+            raise(ValueError('Missing system ID'))
+
+        id_type, entry_id = ids[0]
+
+        if id_type == 'PDB ID':
+            # PDB filenames have to be lowercase
+            entry_id = entry_id.lower()
+        elif id_type == 'PDB-Dev ID':
+            # PDB-Dev filenames have to be uppercase
+            entry_id = entry_id.upper()
+        else:
+            # Use entry ID as is
+            pass
+
         return entry_id
 
     def get_pdb_id(self) -> str:

ihm_validation/report.py

@@ -79,7 +79,9 @@ def run_entry_composition(self, Template_Dict: dict) -> dict:
         Template_Dict['flex'] = utility.get_flex(
             utility.dict_to_JSlist_rows(RB, flex))
         entry_id = self.input.get_id()
+        file_id = self.input.get_file_id()
         Template_Dict['ID'] = entry_id
+        Template_Dict['ID_f'] = file_id
         Template_Dict['PDB_ID'] = self.input.get_pdb_id()
         Template_Dict['PDBDEV_ID'] = self.input.get_pdb_dev_id()
         Template_Dict['ranked_id_list'] = self.input.get_ranked_id_list()

ihm_validation/sas.py

@@ -36,7 +36,7 @@ class SasValidation(GetInputInformation):
     def __init__(self, mmcif_file, db='.'):
         super().__init__(mmcif_file)
         self.version = self.get_atsas_version()
-        self.ID = str(GetInputInformation.get_id(self))
+        self.ID = self.get_id()
         self.nos = GetInputInformation.get_number_of_models(self)
         self.dataset = GetInputInformation.get_dataset_comp(self)
         self.imagepath = '../static/images/'

ihm_validation/sas_plots.py

@@ -19,7 +19,8 @@
 class SasValidationPlots(sas.SasValidation):
     def __init__(self, mmcif_file, imageDirName, driver):
         super().__init__(mmcif_file)
-        self.ID = str(GetInputInformation.get_id(self))
+        self.ID = self.get_id()
+        self.ID_f = self.get_file_id()
         # self.intensities = self.get_intensities()
         # self.intensities = self.modify_intensity()
 
@@ -39,7 +40,7 @@ def plot_intensities(self, sasbdb: str, df: pd.DataFrame):
         '''
         plot intensities with errors
         '''
-        output_file(self.ID+sasbdb+"intensities.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"intensities.html", mode="inline")
         source = ColumnDataSource(df)
         p = figure(plot_height=350, plot_width=350,
                    title=f"Log I(q) vs q with error bars\n{sasbdb}",
@@ -61,16 +62,16 @@ def plot_intensities(self, sasbdb: str, df: pd.DataFrame):
         p.xaxis.axis_label_text_font_style = 'italic'
         p.yaxis.axis_label_text_font_style = 'italic'
 
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"intensities.html")
+        save(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"intensities.html")
         p.output_backend = "svg"
         export_svg(p, height=350, width=350, filename=self.filename +
-                    '/'+self.ID+sasbdb+"intensities.svg", webdriver=self.driver)
+                    '/'+self.ID_f + '_' +sasbdb+"intensities.svg", webdriver=self.driver)
 
     def plot_intensities_log(self, sasbdb: str, df: pd.DataFrame):
         '''
         plot intensities on a log scale with errors
         '''
-        output_file(self.ID+sasbdb+"intensities_log.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"intensities_log.html", mode="inline")
         source = ColumnDataSource(df)
         p = figure(plot_height=350, plot_width=350,
                    title=f"Log I(q) vs Log q with error bars\n{sasbdb}",
@@ -93,16 +94,16 @@ def plot_intensities_log(self, sasbdb: str, df: pd.DataFrame):
         p.yaxis.axis_label_text_font_style = 'italic'
 
         save(p, filename=self.filename+'/' +
-             self.ID+sasbdb+"intensities_log.html")
+             self.ID_f + '_' +sasbdb+"intensities_log.html")
         p.output_backend = "svg"
         export_svg(p, height=350, width=350, filename=self.filename +
-                    '/'+self.ID+sasbdb+"intensities_log.svg", webdriver=self.driver)
+                    '/'+self.ID_f + '_' +sasbdb+"intensities_log.svg", webdriver=self.driver)
 
     def plot_kratky(self, sasbdb: str, df: pd.DataFrame):
         '''
         plot dimensionless kratky
         '''
-        output_file(self.ID+sasbdb+"Kratky.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"Kratky.html", mode="inline")
         source = ColumnDataSource(df)
         p = figure(plot_height=350, plot_width=350,
                    title=f"Dimensionless Kratky plot\n{sasbdb}",
@@ -124,15 +125,15 @@ def plot_kratky(self, sasbdb: str, df: pd.DataFrame):
         p.yaxis.axis_label_text_font_size = '14pt'
         p.xaxis.axis_label_text_font_style = 'italic'
         p.yaxis.axis_label_text_font_style = 'italic'
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"Kratky.html")
+        save(p, filename=self.filename+'/'+self.ID_f+ '_' + sasbdb+"Kratky.html")
         p.output_backend = "svg"
-        export_svg(p, filename=self.filename+'/'+self.ID+sasbdb+"Kratky.svg", webdriver=self.driver)
+        export_svg(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"Kratky.svg", webdriver=self.driver)
 
     def plot_porod_debye(self, sasbdb: str, df: pd.DataFrame):
         '''
         porod debye plot for flexibility
         '''
-        output_file(self.ID+sasbdb+"porod.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"porod.html", mode="inline")
         source = ColumnDataSource(df)
         p = figure(plot_height=350, plot_width=350,
                    title=f"Porod-Debye plot\n{sasbdb}",
@@ -152,14 +153,14 @@ def plot_porod_debye(self, sasbdb: str, df: pd.DataFrame):
         p.xaxis.axis_label_text_font_style = 'italic'
         p.yaxis.axis_label_text_font_style = 'italic'
         p.output_backend = "svg"
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"porod.html")
-        export_svg(p, filename=self.filename+'/'+self.ID+sasbdb+"porod.svg", webdriver=self.driver)
+        save(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"porod.html")
+        export_svg(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"porod.svg", webdriver=self.driver)
 
     def plot_pddf(self, sasbdb: str, df: pd.DataFrame, Rg=None, Dmax=None):
         '''
         p(r) plot, deprecated function
         '''
-        output_file(self.ID+sasbdb+"pddf.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"pddf.html", mode="inline")
         source = ColumnDataSource(df)
         ymax = max(df['P']) * 1.1
 
@@ -197,8 +198,8 @@ def plot_pddf(self, sasbdb: str, df: pd.DataFrame, Rg=None, Dmax=None):
         p.output_backend = "svg"
         p.xaxis.axis_label_text_font_style = 'italic'
         p.yaxis.axis_label_text_font_style = 'italic'
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"pddf.html")
-        export_svg(p, height=350, width=350, filename=self.filename+'/'+self.ID+sasbdb+"pddf.svg", webdriver=self.driver)
+        save(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"pddf.html")
+        export_svg(p, height=350, width=350, filename=self.filename+'/'+self.ID_f+ "_" + sasbdb+"pddf.svg", webdriver=self.driver)
 
     def plot_pddf_int_rwt(self, sasbdb: str,
                       df_int: pd.DataFrame,
@@ -207,7 +208,7 @@ def plot_pddf_int_rwt(self, sasbdb: str,
         '''
         p(r) with fit
         '''
-        output_file(self.ID+sasbdb+"pddf_int.html", mode="inline")
+        output_file(self.ID_f + '_' + sasbdb+"pddf_int.html", mode="inline")
         source1 = ColumnDataSource(df_int)
         source2 = ColumnDataSource(df_pofr)
         p1 = figure(plot_height=350, plot_width=350,
@@ -267,16 +268,16 @@ def plot_pddf_int_rwt(self, sasbdb: str,
         p.width = 350
         p.height = 550
 
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"pddf_int.html")
+        save(p, filename=self.filename+'/'+self.ID_f + '_' + sasbdb+"pddf_int.html")
         # p.output_backend = "svg"
         export_svg(p, filename=self.filename+'/' +
-                    self.ID+sasbdb+"pddf_int.svg", webdriver=self.driver)
+                    self.ID_f + '_' + sasbdb+"pddf_int.svg", webdriver=self.driver)
 
     def Guinier_plot_fit_rwt(self, sasbdb: str, df: pd.DataFrame, score: float):
         '''
         Gunier plot with fit
         '''
-        output_file(self.ID+sasbdb+"guinier.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+"guinier.html", mode="inline")
         source = ColumnDataSource(df)
         p1 = figure(plot_height=350, plot_width=350,
                    title=f"Guinier plot for {sasbdb}")
@@ -338,8 +339,8 @@ def Guinier_plot_fit_rwt(self, sasbdb: str, df: pd.DataFrame, score: float):
         p.height = 550
 
 
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+"guinier.html")
-        export_svg(p, filename=self.filename+'/'+self.ID+sasbdb+"guinier.svg", webdriver=self.driver)
+        save(p, filename=self.filename+'/'+self.ID_f+ '_' + sasbdb+"guinier.html")
+        export_svg(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+"guinier.svg", webdriver=self.driver)
 
 
     def plot_multiple(self):
@@ -384,7 +385,7 @@ def plot_fit_rwt(self, sasbdb: str, fit: int, score: float, df: pd.DataFrame):
         '''
         plot chi-squared fit
         '''
-        output_file(self.ID+sasbdb+str(fit)+"fit1.html", mode="inline")
+        output_file(self.ID_f + '_' +sasbdb+str(fit)+"fit1.html", mode="inline")
         source = ColumnDataSource(df)
         p1 = figure(plot_height=350, plot_width=350,
                    title=f"Model fit for {sasbdb}, model {fit + 1}")
@@ -440,7 +441,7 @@ def plot_fit_rwt(self, sasbdb: str, fit: int, score: float, df: pd.DataFrame):
         p = gridplot([p1, p2], ncols=1,
                      toolbar_location="above")
 
-        save(p, filename=self.filename+'/'+self.ID+sasbdb+str(fit)+"fit1.html")
+        save(p, filename=self.filename+'/'+self.ID_f + '_' +sasbdb+str(fit)+"fit1.html")
         # p.output_backend = "svg"
         export_svg(p, filename=self.filename+'/' +
-                    self.ID+sasbdb+str(fit)+"fit1.svg", webdriver=self.driver)
+                    self.ID_f + '_' +sasbdb+str(fit)+"fit1.svg", webdriver=self.driver)

templates/data_quality.html

@@ -60,9 +60,9 @@ <h5 class= ex2 align= center>
               <!-- plots -->
               <div class=row>
                 <!-- log I(q) vs q plot -->
-                {{ insert_sas_plot(ID, sasdb_code_html[i], "intensities") }}
+                {{ insert_sas_plot(ID_f, sasdb_code_html[i], "intensities") }}
                 <!-- log I(q) vs log q plot -->
-                {{ insert_sas_plot(ID, sasdb_code_html[i], "intensities_log") }}
+                {{ insert_sas_plot(ID_f, sasdb_code_html[i], "intensities_log") }}
               </div>
             </div>
           </div>
@@ -112,9 +112,9 @@ <h5 class= ex2 align= center>
           <!-- plots -->
           <div class=row>
             <!-- Porod-Debye plot -->
-            {{ insert_sas_plot(ID, sasdb_code_html[i], "porod") }}
+            {{ insert_sas_plot(ID_f, sasdb_code_html[i], "porod") }}
             <!-- Dimensionless Kratky plot -->
-            {{ insert_sas_plot(ID, sasdb_code_html[i], "Kratky") }}
+            {{ insert_sas_plot(ID_f, sasdb_code_html[i], "Kratky") }}
           </div>
         {% endfor %}
       </div>
@@ -142,9 +142,9 @@ <h5 align= center>
           <!-- main plots -->
           <div class=row>
             <!-- Pair distance distribution function -->
-            {{ insert_sas_plot(ID, sasdb_code_html[i], "pddf") }}
+            {{ insert_sas_plot(ID_f, sasdb_code_html[i], "pddf") }}
             <!-- P(r) extrapolated fit -->
-            {{ insert_sas_plot(ID, sasdb_code_html[i], "pddf_int") }}
+            {{ insert_sas_plot(ID_f, sasdb_code_html[i], "pddf_int") }}
           </div>
         {% endfor %}
       </div>
@@ -183,7 +183,7 @@ <h5 class= ex2 align= center><u><a name=firstG> Guinier analysis </a></u><a clas
             <!-- Guinier plot -->
             <div class=row>
               {% for i in range(sasdb_code_html|length) %}
-                {{ insert_sas_plot(ID, sasdb_code_html[i], "guinier") }}
+                {{ insert_sas_plot(ID_f, sasdb_code_html[i], "guinier") }}
               {% endfor %}
             </div>
           </div>

templates/formodeling.html

@@ -56,7 +56,7 @@ <h5 class= ex2 align= center>
                 {% for i in range(sasdb_code_html|length) %}
                   {% for j in range(sasdb_code_fits[i]) %}
                     {% set k = j + 1 %}
-                      {{ insert_sas_plot(ID, sasdb_code_html[i], j|int ~ 'fit1') }}
+                      {{ insert_sas_plot(ID_f, sasdb_code_html[i], j|int ~ 'fit1') }}
                   {% endfor %}
                 {% endfor %}
               </div>