Update src/streamlit_app.py

src/streamlit_app.py

@@ -28,25 +28,138 @@ import subprocess
 import requests
 from transformers import pipeline, BioGptTokenizer, BioGptForCausalLM
 from transformers import MarianMTModel, MarianTokenizer
+import logging
+
+# Configure logging
+logging.basicConfig(level=logging.DEBUG)
+logger = logging.getLogger(__name__)
 
 # Set page configuration as the first Streamlit command
 st.set_page_config(page_title="Omics Risk Predictor", layout="wide", initial_sidebar_state="expanded")
 
+# Set Hugging Face cache
+os.environ["HF_HOME"] = "/app/hf_cache"
+os.makedirs("/app/hf_cache", exist_ok=True)
+
 # Load translations without Streamlit commands
 def load_translations_raw():
     try:
+        logger.debug(f"Current directory: {os.getcwd()}")
+        logger.debug(f"Translations file path: {os.path.abspath('src/translations.json')}")
         with open("src/translations.json", "r", encoding="utf-8") as f:
-            return json.load(f)
-    except FileNotFoundError:
-        return {}
-    except json.JSONDecodeError:
-        return {}
+            data = json.load(f)
+            logger.debug("Translations loaded successfully")
+            return data
+    except FileNotFoundError as e:
+        logger.error(f"Translation file not found: {e}")
+        return {
+            "en": {
+                "app_title": "Omics Risk Predictor",
+                "patient_id": "Patient ID",
+                "select_language": "Select Language",
+                "home": "Home",
+                "about_omics": "About Omics Data",
+                "predict_risk": "Predict Risk",
+                "chatbot": "Chatbot Assistant",
+                "dashboard": "Dashboard",
+                "footer": "© 2025, Ngoue David Roger Yannick",
+                "upload_warning": "Please upload all omics files and provide a patient ID.",
+                "risk_score": "Risk Score",
+                "ckd_risk_score": "CKD Risk Score: {score:.2f}%",
+                "key_biomarkers": "Key Biomarkers",
+                "download_results": "Download Results",
+                "chatbot_title": "Chatbot Assistant",
+                "chatbot_intro": "Interact with our assistant for recommendations.",
+                "chat_with_assistant": "Chat with Assistant",
+                "type_message": "Type your message",
+                "send": "Send",
+                "biogpt_load_error": "Error loading BioGPT",
+                "recommendation_error": "Error generating recommendation",
+                "medication_error": "Error generating medication recommendation",
+                "diet_error": "Error generating diet recommendation",
+                "exercise_error": "Error generating exercise recommendation",
+                "css_error": "CSS file not found",
+                "model_weights_error": "Model weights not found",
+                "model_load_error": "Error loading model: {error}",
+                "file_process_error": "Error processing files: {error}",
+                "chat_risk_keywords": "risk,score",
+                "chat_medication_keywords": "medication,drug",
+                "chat_diet_keywords": "diet,food",
+                "chat_exercise_keywords": "exercise,workout",
+                "chat_risk_response": "Risk score: {score:.2f}%. Biomarkers: {biomarkers}.",
+                "shap_top_features": "Top SHAP Features",
+                "shap_mean_value": "SHAP Mean Value",
+                "feature": "Feature",
+                "omic": "Omic",
+                "omics": "Omics",
+                "genomics": "Genomics",
+                "transcriptomics": "Transcriptomics",
+                "proteomics": "Proteomics",
+                "metabolomics": "Metabolomics",
+                "recommendations": "Recommendations",
+                "download_biomarkers": "Download Biomarkers",
+                "download_report": "Download Report",
+                "shap": "SHAP Analysis"
+            }
+        }
+    except json.JSONDecodeError as e:
+        logger.error(f"JSON decode error: {e}")
+        return {
+            "en": {
+                "app_title": "Omics Risk Predictor",
+                "patient_id": "Patient ID",
+                "select_language": "Select Language",
+                "home": "Home",
+                "about_omics": "About Omics Data",
+                "predict_risk": "Predict Risk",
+                "chatbot": "Chatbot Assistant",
+                "dashboard": "Dashboard",
+                "footer": "© 2025, Ngoue David Roger Yannick",
+                "upload_warning": "Please upload all omics files and provide a patient ID.",
+                "risk_score": "Risk Score",
+                "ckd_risk_score": "CKD Risk Score: {score:.2f}%",
+                "key_biomarkers": "Key Biomarkers",
+                "download_results": "Download Results",
+                "chatbot_title": "Chatbot Assistant",
+                "chatbot_intro": "Interact with our assistant for recommendations.",
+                "chat_with_assistant": "Chat with Assistant",
+                "type_message": "Type your message",
+                "send": "Send",
+                "biogpt_load_error": "Error loading BioGPT",
+                "recommendation_error": "Error generating recommendation",
+                "medication_error": "Error generating medication recommendation",
+                "diet_error": "Error generating diet recommendation",
+                "exercise_error": "Error generating exercise recommendation",
+                "css_error": "CSS file not found",
+                "model_weights_error": "Model weights not found",
+                "model_load_error": "Error loading model: {error}",
+                "file_process_error": "Error processing files: {error}",
+                "chat_risk_keywords": "risk,score",
+                "chat_medication_keywords": "medication,drug",
+                "chat_diet_keywords": "diet,food",
+                "chat_exercise_keywords": "exercise,workout",
+                "chat_risk_response": "Risk score: {score:.2f}%. Biomarkers: {biomarkers}.",
+                "shap_top_features": "Top SHAP Features",
+                "shap_mean_value": "SHAP Mean Value",
+                "feature": "Feature",
+                "omic": "Omic",
+                "omics": "Omics",
+                "genomics": "Genomics",
+                "transcriptomics": "Transcriptomics",
+                "proteomics": "Proteomics",
+                "metabolomics": "Metabolomics",
+                "recommendations": "Recommendations",
+                "download_biomarkers": "Download Biomarkers",
+                "download_report": "Download Report",
+                "shap": "SHAP Analysis"
+            }
+        }
 
 translations = load_translations_raw()
 
 # Get translation function
 def t(key, lang="fr"):
-    return translations.get(lang, {}).get(key, key)
+    return translations.get(lang, {}).get(key, translations.get("en", {}).get(key, key))
 
 # Check for translation file errors after set_page_config
 if not translations:
@@ -93,7 +206,7 @@ Session = sessionmaker(bind=engine)
 # Sidebar
 with st.sidebar:
     st.markdown(f"<div class='sidebar-header'>{t('app_title')}</div>", unsafe_allow_html=True)
-    st.image("https://via.placeholder.com/150", caption=t("author"))
+    st.image("https://via.placeholder.com/150", caption="Ngoue David Roger Yannick")
     st.markdown(f"<div class='sidebar-subheader'>{t('navigation')}</div>", unsafe_allow_html=True)
 
     # Language selector
@@ -107,6 +220,7 @@ with st.sidebar:
 
 # Device configuration
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+logger.debug(f"Using device: {device}")
 
 # IRC biomarkers
 irc_biomarkers = [
@@ -327,9 +441,9 @@ def generate_pdf_report(risk_score, important_biomarkers, shap_plot_path, recomm
     try:
         c = canvas.Canvas(output_path, pagesize=letter)
         c.setFont("Helvetica-Bold", 16)
-        c.drawString(100, 750, t('report_title', lang))
+        c.drawString(100, 750, t('app_title', lang))
         c.setFont("Helvetica", 12)
-        c.drawString(100, 730, t('author', lang))
+        c.drawString(100, 730, "Ngoue David Roger Yannick")
         c.drawString(100, 710, t('footer', lang))
 
         c.setFont("Helvetica-Bold", 14)
@@ -362,13 +476,19 @@ def generate_pdf_report(risk_score, important_biomarkers, shap_plot_path, recomm
         raise Exception(t("pdf_error").format(error=str(e)))
 
 # Load BioGPT for recommendations
+biogpt_tokenizer = None
+biogpt_model = None
 try:
+    logger.debug("Loading BioGPT tokenizer...")
     biogpt_tokenizer = BioGptTokenizer.from_pretrained("microsoft/biogpt")
+    logger.debug("Loading BioGPT model...")
     biogpt_model = BioGptForCausalLM.from_pretrained("microsoft/biogpt").to(device)
+    logger.debug("Setting BioGPT model to eval mode...")
     biogpt_model.eval()
+    logger.debug("BioGPT loaded successfully")
 except Exception as e:
-    st.error(t("biogpt_load_error"))
-    st.stop()
+    logger.error(f"Failed to load BioGPT: {str(e)}")
+    st.warning(t("biogpt_load_error") + f": {str(e)}. Using mock recommendations.")
 
 # Load translation model (English to French)
 try:
@@ -381,6 +501,8 @@ except Exception as e:
 # Generate recommendation using BioGPT
 @st.cache_data
 def generate_recommendation(prompt, max_length=100):
+    if biogpt_model is None:
+        return t("recommendation_error") + " (BioGPT not loaded, using mock response)"
     try:
         inputs = biogpt_tokenizer(prompt, return_tensors="pt", max_length=512, truncation=True).to(device)
         outputs = biogpt_model.generate(
@@ -515,7 +637,7 @@ elif page == t("about_omics"):
             <li><b>{t('genomics')}</b>: {t('genomics_format')}</li>
             <li><b>{t('transcriptomics')}</b>: {t('transcriptomics_format')}</li>
             <li><b>{t('proteomics')}</b>: {t('proteomics_format')}</li>
-            <li><b>{t('metabolomics')}</b>: {t('metabolomics_format')}</li>
+            <li><b>{t('metabolomics')}</b></li>
         </ul>
         <p>{t('raw_omics_processing')}</p>
     </div>
@@ -527,52 +649,52 @@ elif page == t("about_omics"):
         <p>{t('preprocessing_desc')}</p>
         <ol>
             <li><b>{t('quality_control')}</b>: {t('quality_control_desc')}</li>
-            <li><b>{t('normalization')}</b>: {t('normalization_desc')}</li>
-            <li><b>{t('imputation')}</b>: {t('imputation_desc')}</li>
-            <li><b>{t('standardization')}</b>: {t('standardization_desc')}</li>
+            <li><b>{t('normalization')}</b></li>
+            <li><b>{t('imputation')}</b>: {t('imputation_desc')}</b></li>
+            <li><b>{t('standardization')}</b></li>
             <li><b>{t('feature_selection')}</b>: {t('feature_selection_desc')}</li>
         </ol>
         <p>{t('preprocessing_importance')}</p>
     </div>
     """, unsafe_allow_html=True)
 
-    st.markdown(f"<div class='section-header'>{t('csv_format')}</div>", unsafe_allow_html=True)
-    st.markdown(f"""
+    st.markdown(f"<div class='section-title'>{t('csv_format')}</div>", unsafe_allow_html=True)
+    st.markdown("""
     <div class='content animate__fadeIn'>
         <p>{t('csv_format_desc')}</p>
         <ul>
             <li><b>{t('columns')}</b>: {t('columns_desc')}</li>
-            <li><b>{t('rows')}</b>: {t('rows_desc')}</li>
+            <li><b>{t('rows')}</b>: {t('rows_desc')}</p>
             <li><b>{t('values')}</b>: {t('values_desc')}</li>
-            <li><b>{t('separator')}</b>: {t('separator_desc')}</li>
-            <li><b>{t('header')}</b>: {t('header_desc')}</li>
+            <li><b>{t('separator')}</b></li>
+            </li><b>{t('header')}</b>: {t('header_desc')}</p>
         </ul>
         <p>{t('csv_example')}</p>
         <pre class='code-block'>
-PatientID,UMOD_rs12917707,APOL1_rs73885319,MYH9_rs4821480,Status
-P001,0,1,0,Unknown
+            Patient_ID,UMOD_rs12917707,APOL1_rs73885319,MYH9_rs4821480
+            P001,0,0,1
         </pre>
         <p>{t('csv_patient_id_desc')}</p>
     </div>
     """, unsafe_allow_html=True)
 
-    st.markdown(f"<div class='section-header'>{t('data_quality')}</div>", unsafe_allow_html=True)
+    st.markdown(f"<div class='section-title'>{t('data_quality')}</div>", unsafe_allow_html=True)
     st.markdown(f"""
     <div class='content animate__fadeIn'>
         <p>{t('data_quality_desc')}</p>
         <ul>
-            <li><b>{t('consistency')}</b>: {t('consistency_desc')}</li>
+            <li><b>{t('consistency')}</b></li>
             <li><b>{t('completeness')}</b>: {t('completeness_desc')}</li>
-            <li><b>{t('accuracy')}</b>: {t('accuracy_desc')}</li>
-            <li><b>{t('alignment')}</b>: {t('alignment_desc')}</li>
+            <li><b>{t('accuracy')}</b>: {t('accuracy_desc')}</b>
+            <li><b>{t('alignment')}</b>: {t('alignment_desc')}</b></li>
         </ul>
-        <p>{t('data_quality_importance')}</p>
+        <p>{t('data_quality_desc')}</p>
     </div>
     """, unsafe_allow_html=True)
 
 # Predict Risk Page
 elif page == t("predict_risk"):
-    st.markdown(f"<div class='title animate__fadeIn'>{t('predict_risk_title')}</div>", unsafe_allow_html=True)
+    st.markdown(f"<div class='title animate__fadeIn'>{t('predict_risk')}</div>", unsafe_allow_html=True)
     st.markdown(f"""
     <div class='content animate__fadeIn'>
         <p>{t('predict_risk_intro')}</p>
@@ -607,7 +729,7 @@ elif page == t("predict_risk"):
 
     # Validate and process files
     if st.button(t("run_prediction"), key="run_prediction"):
-        if all(file_paths[omic] for omic in file_paths) and st.session_state.patient_id:
+        if all(file_paths.values()) and st.session_state.patient_id:
             try:
                 with st.spinner(t("processing")):
                     # Load and preprocess data
@@ -620,7 +742,7 @@ elif page == t("predict_risk"):
                     patient_tensors = [torch.tensor(data_dict[omic].values, dtype=torch.float32) for omic in data_dict]
                     patient_dataset = TensorDataset(*patient_tensors)
                     patient_loader = DataLoader(patient_dataset, batch_size=best_hyperparams['batch_size'],
-                                               shuffle=False)
+                                   shuffle=False)
 
                     # Extract latent representations
                     encoded_data = extract_latent_representations(model, patient_loader, device)
@@ -666,7 +788,7 @@ elif page == t("predict_risk"):
                     # Store results
                     st.session_state.results = {
                         'umap_df': umap_df,
-                        'important_biomarkers': important_biomarkers,
+                        'important_biom': important_biomarkers,
                         'output_dir': output_dir,
                         'pdf_path': pdf_path,
                         'recommendations': recommendations
@@ -674,8 +796,8 @@ elif page == t("predict_risk"):
 
                     # Display results
                     st.markdown(f"<div class='section-header'>{t('prediction_results')}</div>", unsafe_allow_html=True)
-                    st.markdown(f"<div class='result-box'>{t('ckd_risk_score').format(score=risk_score):.2f}</div>",
-                                unsafe_allow_html=True)
+                    st.markdown(f"<div class='result-box'>{t('ckd_risk_score').format(score=risk_score):,.2f}</div>",
+                            unsafe_allow_html=True)
 
                     # Risk gauge
                     fig_gauge = go.Figure(go.Indicator(
@@ -686,7 +808,7 @@ elif page == t("predict_risk"):
                             'axis': {'range': [0, 100], 'tickcolor': '#FFF', 'tickfont': {'color': '#FFF'}},
                             'bar': {'color': '#0000FF'},
                             'steps': [
-                                {'range': [0, 33], 'color': '#4ADE80'},
+                                {'range': [0, 33], 'color': '#FFD700'},
                                 {'range': [33, 66], 'color': '#FACC15'},
                                 {'range': [66, 100], 'color': '#EF4444'}
                             ],
@@ -699,7 +821,7 @@ elif page == t("predict_risk"):
                     ))
                     fig_gauge.update_layout(
                         paper_bgcolor='#0E1117',
-                        font={'color': "#FFFFFF"},
+                        font={'color': '#FFFFFF'},
                         margin={'l': 20, 'r': 20, 't': 50, 'b': 20}
                     )
                     st.plotly_chart(fig_gauge, use_container_width=True)
@@ -708,7 +830,7 @@ elif page == t("predict_risk"):
                     st.markdown(f"<div class='subheader'>{t('key_biomarkers')}</div>", unsafe_allow_html=True)
                     shap_fig = plt.figure(figsize=(12, 8))
                     sns.barplot(data=important_biomarkers.head(20), x='SHAP_Mean_Abs', y=t('feature'), hue=t('omic'),
-                                palette=['#FF6F61', '#6B5B95', '#4CAF50', '#2196F3'])
+                                    palette=['#FF6F61', '#6B5B95', '#4CAF50', '#2196F3'])
                     plt.title(t('shap_top_features'), fontsize=16, color='#FFFFFF')
                     plt.xlabel(t('shap_mean_value'), fontsize=12, color='#FFFFFF')
                     plt.ylabel(t('feature'), fontsize=12, color='#FFFFFF')
@@ -720,26 +842,26 @@ elif page == t("predict_risk"):
 
                     # Display saved SHAP plot
                     if os.path.exists(shap_plot_path):
-                        st.image(shap_plot_path, caption=t('shap'), use_column_width=True)
+                        st.image(shap_plot_path, caption=t('shap'), use_column_width=True
 
                     # Recommendations
                     st.markdown(f"<div class='subheader'>{t('recommendations')}</div>", unsafe_allow_html=True)
                     for rec in recommendations:
-                        st.markdown(f"<div class='recommendation'>{rec}</div>", unsafe_allow_html=True)
+                        st.markdown(f"<div class='recommendation'>{rec['text']}</div>", unsafe_allow_html=True)
 
                     # Download results
                     st.markdown(f"<div class='subheader'>{t('download_results')}</div>", unsafe_allow_html=True)
                     col1, col2 = st.columns(2)
                     with col1:
-                        csv = important_biomarkers.to_csv(index=False).encode('utf-8')
+                        csv = important_biom.to_csv(index=False).encode('utf-8')
                         b64 = base64.b64encode(csv).decode()
-                        href = f'<a href="data:file/csv;base64,{b64}" download="important_biomarkers.csv" class="download-btn">{t("download_biomarkers", lang)}</a>'
-                        st.markdown(href, unsafe_allow_html=True)
+                        href = f'<a href="data:///csv;base64,{b64}" download="important_biomarkers.csv" class="download-btn">{t("download_biomarkers", lang)}</a>'
+                        st.markdown(f"{href, unsafe_allow_html=True)
                     with col2:
                         with open(pdf_path, "rb") as f:
                             pdf_data = f.read()
                         b64_pdf = base64.b64encode(pdf_data).decode()
-                        href_pdf = f'<a href="data:application/pdf;base64,{b64_pdf}" download="ckd_risk_report.pdf" class="download-btn">{t("download_report", lang)}</a>'
+                        href_pdf = f'<a href="data:application/pdf;base64,b64_pdf}" download="ckd_risk_report.pdf" class="download-btn">{t("download_report", lang)}</a>'
                         st.markdown(href_pdf, unsafe_allow_html=True)
 
             except Exception as e:
@@ -753,7 +875,7 @@ elif page == t("chatbot"):
     st.markdown(f"<div class='title animate__fadeIn'>{t('chatbot_title')}</div>", unsafe_allow_html=True)
     st.markdown(f"""
     <div class='content animate__fadeIn'>
-        <p>{t('chatbot_intro')}</p>
+        <p>{t('chatbot_int')}</p>
     </div>
     """, unsafe_allow_html=True)
 
@@ -768,18 +890,18 @@ elif page == t("chatbot"):
                 st.markdown(f"<div class='chat-bubble bot'>{message['content']}</div>", unsafe_allow_html=True)
 
     # User input
-    user_input = st.text_area(t("type_message"), key="chat_input")
-    if st.button(t("send"), key="button_send"):
+    user_input = st.text_area(t("type_message"), key="chat_input"))
+    if st.button(t("send"), key="button_send")):
         if user_input:
             try:
-                # Add user message to history
+                # Append user message to history
                 st.session_state.chat_history.append({'role': 'user', 'content': user_input})
 
                 # Check if user is asking about results
                 if st.session_state.results and any(
                         keyword in user_input.lower() for keyword in t('chat_risk_keywords').split(',')):
                     risk_score = st.session_state.results['umap_df'][t('risk_score')].iloc[0]
-                    biomarkers = st.session_state.results['important_biomarkers'][t('feature')].head(5).tolist()
+                    biomarkers = st.session_state.results['important_biom'][t('feature')].head(5).tolist()
                     context = t('chat_risk_response').format(score=risk_score, biomarkers=', '.join(biomarkers))
                     response = send_to_rasa(f"{context} {user_input}", lang)
                 else:
@@ -794,7 +916,7 @@ elif page == t("chatbot"):
 
                     if intent != 'None' and st.session_state.results:
                         risk_score = st.session_state.results['umap_df'][t('risk_score')].iloc[0]
-                        biomarkers = st.session_state.results['important_biomarkers'][t('feature')].head(5).tolist()
+                        biomarkers = st.session_state.results['important_biom'][t('feature')].head(5).tolist()
                         prompt = formulate_prompt(intent, risk_score, biomarkers)
                         response = generate_recommendation(prompt)
                         if validate_recommendation(response):
@@ -804,11 +926,12 @@ elif page == t("chatbot"):
                     else:
                         response = send_to_rasa(user_input, lang)
 
-                # Add bot response to history
+                # Append bot response to history
                 st.session_state.chat_history.append({'role': 'bot', 'content': response})
 
                 # Rerender chat
                 st.rerun()
+            )
 
             except Exception as e:
                 st.error(t("chatbot_error").format(error=str(e)))
@@ -818,7 +941,7 @@ elif page == t("dashboard"):
     st.markdown(f"<div class='title animate__fadeIn'>{t('dashboard_title')}</div>", unsafe_allow_html=True)
     st.markdown(f"""
     <div class='content animate__fadeIn'>
-        <p>{t('dashboard_intro')}</p>
+        <p>{t('dashboard_int')}</p>
     </div>
     """, unsafe_allow_html=True)
 
@@ -834,17 +957,18 @@ elif page == t("dashboard"):
             } for p in patients]
             df_patients = pd.DataFrame(patient_data)
 
-            # Filters
-            st.markdown(f"<div class='subheader'>{t('filters')}</div>", unsafe_allow_html=True)
-            risk_threshold = st.slider(t("risk_threshold"), 0, 100, 50)
+            # Filter data
+            st.markdown(f"<div class='section-title'>{t('filters')}</div>", unsafe_allow_html=True)
+            risk_threshold = t('risk_threshold')
+            st.markdown(f"Slider(t({risk_threshold}), 0, 100, 50)")
             filtered_df = df_patients[df_patients[t('risk_score')] >= risk_threshold]
 
             # Display table
-            st.dataframe(filtered_df, use_container_width=True)
+            st.markdown(filtered_df, use_container_width=True)
 
             # Plot
             fig = px.scatter(
-                filtered_df,
+                data=filtered_df,
                 x=t('risk_score'),
                 y=t('patient_id'),
                 color=t('risk_score'),
@@ -874,7 +998,7 @@ if rasa_process:
     try:
         rasa_process.terminate()
     except Exception as e:
-        print(f"Error terminating Rasa server: {str(e)}")
+        print(f"Error occurred terminating Rasa server: {str(e)}")
 
 st.markdown(f"""
 <div class='footer'>