Update app.py

app.py

@@ -7,36 +7,25 @@ from io import StringIO, BytesIO
 
 st.set_page_config(page_title="PCR Primer Designer", layout="wide")
 
-st.sidebar.title("Instructions")
-st.sidebar.info("""
-1. Upload a GenBank (.gb/.gbk) file.
-2. Choose a CDS from the dropdown (length shown).
-3. Set desired PCR product size range.
-4. Design primers and visualize binding.
-5. Optionally filter similar-sized products.
-""")
-
-uploaded_file = st.file_uploader("Upload GenBank file", type=["gb", "gbk"])
-
+uploaded_file = st.file_uploader("Upload GenBank File", type=["gb", "gbk"])
 plot_export_format = st.selectbox("Export Plot Format", ["PNG", "PDF", "SVG"])
 
-def extract_features(record):
-    features = []
-    for f in record.features:
-        if f.type == "CDS":
-            gene = f.qualifiers.get("gene", ["unnamed"])[0]
-            loc = f.location
-            length = len(f.location)
-            features.append((gene, loc, length, f))
-    return features
+def extract_cds_features(record):
+    return [
+        (
+            f.qualifiers.get("gene", ["unnamed"])[0],
+            f.location,
+            len(f.location),
+            f
+        )
+        for f in record.features if f.type == "CDS"
+    ]
 
 def design_primers(seq, size_range, count):
     min_size, max_size = size_range
     return primer3.bindings.designPrimers(
-        {
-            'SEQUENCE_TEMPLATE': str(seq),
-            'PRIMER_PRODUCT_SIZE_RANGE': [[min_size, max_size]]
-        },
+        {'SEQUENCE_TEMPLATE': str(seq),
+         'PRIMER_PRODUCT_SIZE_RANGE': [[min_size, max_size]]},
         {
             'PRIMER_OPT_SIZE': 20,
             'PRIMER_MIN_SIZE': 18,
@@ -46,7 +35,7 @@ def design_primers(seq, size_range, count):
             'PRIMER_MAX_TM': 63.0,
             'PRIMER_MIN_GC': 20.0,
             'PRIMER_MAX_GC': 80.0,
-            'PRIMER_NUM_RETURN': count * 2
+            'PRIMER_NUM_RETURN': count
         }
     )
 
@@ -65,100 +54,57 @@ def plot_binding(seq_len, l_start, l_len, r_start, r_len, product_len):
     ax.set_title("Primer Binding Sites and PCR Product", fontsize=14)
     return fig
 
-def is_unique(primer_set, seen_seqs):
-    l_seq = primer_set.get('left')
-    r_seq = primer_set.get('right')
-    if (l_seq, r_seq) in seen_seqs:
-        return False
-    seen_seqs.add((l_seq, r_seq))
-    return True
-
-def filter_similar_by_product(primers, threshold=5):
-    seen_sizes = set()
-    filtered = []
-    for row in primers:
-        size = row['PCR Product Size']
-        if any(abs(size - s) <= threshold for s in seen_sizes):
-            continue
-        seen_sizes.add(size)
-        filtered.append(row)
-    return filtered
-
 if uploaded_file:
     record = SeqIO.read(StringIO(uploaded_file.getvalue().decode("utf-8")), "genbank")
-    cds_features = extract_features(record)
+    cds_list = extract_cds_features(record)
 
-    if cds_features:
+    if cds_list:
         st.subheader("Select CDS Feature")
-        cds_labels = [f"{g} ({loc.start}:{loc.end}) [length: {length} bp]" for g, loc, length, _ in cds_features]
-        selected_idx = st.selectbox("Choose CDS:", range(len(cds_labels)), format_func=lambda x: cds_labels[x])
-        selected_feature = cds_features[selected_idx]
-        _, _, cds_length, feature = selected_feature
-        feature_seq = feature.extract(record.seq)
+        cds_labels = [f"{g} ({loc.start}:{loc.end}) [length: {length} bp]" for g, loc, length, _ in cds_list]
+        selected_idx = st.selectbox("Choose CDS:", range(len(cds_labels)), format_func=lambda i: cds_labels[i])
+        _, _, cds_length, selected_feature = cds_list[selected_idx]
+        seq = selected_feature.extract(record.seq)
 
-        st.code(str(feature_seq), language="text")
+        st.code(str(seq), language="text")
 
         st.subheader("Primer Design Parameters")
-        min_range = 100
-        max_range = cds_length
-        product_range = st.slider("PCR Product Size Range (bp)", min_value=min_range, max_value=max_range,
-                                  value=(min_range, max_range))
-
-        num_primers = st.number_input("Number of primer pairs to return:", min_value=1, value=5, step=1)
-        filter_checkbox = st.checkbox("Filter similar product sizes (±5 bp)", value=False)
+        min_size = 100
+        product_range = st.slider("PCR Product Size Range", min_value=min_size, max_value=cds_length,
+                                  value=(min_size, cds_length))
+        primer_count = st.number_input("Number of Primer Pairs to Return", min_value=1, value=5, step=1)
 
         if st.button("Design Primers"):
             with st.spinner("Designing primers..."):
-                result = design_primers(feature_seq, product_range, num_primers)
-                seen_seqs = set()
-                primer_rows = []
-
-                for i in range(result.get('PRIMER_PAIR_NUM_RETURN', 0)):
-                    l_seq = result.get(f'PRIMER_LEFT_{i}_SEQUENCE')
-                    r_seq = result.get(f'PRIMER_RIGHT_{i}_SEQUENCE')
-                    l_pos, l_len = result.get(f'PRIMER_LEFT_{i}', [None, None])
-                    r_pos, r_len = result.get(f'PRIMER_RIGHT_{i}', [None, None])
-                    prod_size = result.get(f'PRIMER_PAIR_{i}_PRODUCT_SIZE')
-                    if None in (l_seq, r_seq, l_pos, l_len, r_pos, r_len, prod_size):
-                        continue
-                    if is_unique({'left': l_seq, 'right': r_seq}, seen_seqs):
-                        primer_rows.append({
-                            "Primer Pair": i + 1,
-                            "Left Primer": l_seq,
-                            "Right Primer": r_seq,
-                            "PCR Product Size": prod_size,
-                            "Left TM (°C)": round(result.get(f'PRIMER_LEFT_{i}_TM', 0), 2),
-                            "Right TM (°C)": round(result.get(f'PRIMER_RIGHT_{i}_TM', 0), 2),
-                            "Left Pos": l_pos,
-                            "Left Len": l_len,
-                            "Right Pos": r_pos,
-                            "Right Len": r_len,
-                        })
-
-                if filter_checkbox:
-                    primer_rows = filter_similar_by_product(primer_rows)
-
-                if primer_rows:
-                    st.subheader("Designed Primers")
-                    df = pd.DataFrame(primer_rows).drop(columns=["Left Pos", "Left Len", "Right Pos", "Right Len"])
-                    st.dataframe(df)
-
-                    csv = df.to_csv(index=False).encode('utf-8')
-                    st.download_button("Download Primer Table as CSV", csv, "primers.csv", "text/csv")
-
-                    # Plot first pair
-                    first = primer_rows[0]
-                    fig = plot_binding(len(feature_seq), first["Left Pos"], first["Left Len"],
-                                       first["Right Pos"], first["Right Len"], first["PCR Product Size"])
-                    st.pyplot(fig)
-
-                    # Export image
-                    buf = BytesIO()
-                    fmt = plot_export_format.lower()
-                    fig.savefig(buf, format=fmt, dpi=300)
-                    st.download_button(f"Download Binding Plot ({plot_export_format})",
-                                       data=buf.getvalue(),
-                                       file_name=f"primer_plot.{fmt}",
-                                       mime=f"image/{'svg+xml' if fmt == 'svg' else fmt}")
-                else:
-                    st.warning("No unique primers found.")
+                result = design_primers(seq, product_range, primer_count)
+
+                data = []
+                for i in range(result['PRIMER_PAIR_NUM_RETURN']):
+                    data.append({
+                        "Primer Pair": i + 1,
+                        "Left Primer": result.get(f'PRIMER_LEFT_{i}_SEQUENCE'),
+                        "Right Primer": result.get(f'PRIMER_RIGHT_{i}_SEQUENCE'),
+                        "PCR Product Size": result.get(f'PRIMER_PAIR_{i}_PRODUCT_SIZE'),
+                        "Left TM (°C)": round(result.get(f'PRIMER_LEFT_{i}_TM', 0), 2),
+                        "Right TM (°C)": round(result.get(f'PRIMER_RIGHT_{i}_TM', 0), 2)
+                    })
+
+                st.subheader("Designed Primers")
+                df = pd.DataFrame(data)
+                st.dataframe(df)
+
+                csv = df.to_csv(index=False).encode('utf-8')
+                st.download_button("Download Primer Table", csv, "primers.csv", "text/csv")
+
+                # Visualize first primer pair
+                left = result['PRIMER_LEFT_0']
+                right = result['PRIMER_RIGHT_0']
+                size = result['PRIMER_PAIR_0_PRODUCT_SIZE']
+                fig = plot_binding(len(seq), left[0], left[1], right[0], right[1], size)
+                st.pyplot(fig)
+
+                # Export plot
+                buf = BytesIO()
+                ext = plot_export_format.lower()
+                fig.savefig(buf, format=ext, dpi=300)
+                st.download_button(f"Download Plot ({plot_export_format})", buf.getvalue(),
+                                   f"primer_plot.{ext}", mime=f"image/{'svg+xml' if ext == 'svg' else ext}")