Spaces:

Eitan177
/

InSilicoControl

Running

App Files Files Community

Halper-Stromberg commited on Jun 3

Commit

b87cabb

1 Parent(s): 38cfbbb

Update Dockerfile to copy .streamlit config and update src/ files with customized app code

Browse files

Files changed (3) hide show

Dockerfile +1 -0
src/pipeline.py +1 -1
src/streamlit_app.py +334 -2

Dockerfile CHANGED Viewed

@@ -12,6 +12,7 @@ RUN apt-get update && apt-get install -y \
 COPY requirements.txt ./
 COPY src/ ./src/
 RUN pip3 install -r requirements.txt

 COPY requirements.txt ./
 COPY src/ ./src/
+COPY .streamlit/ ./.streamlit/
 RUN pip3 install -r requirements.txt

src/pipeline.py CHANGED Viewed

@@ -448,4 +448,4 @@ def generate_synthetic_bam(
     output_bam.unlink(missing_ok=True)
     log("BAM and VCF generation complete.", log_func)
-    return sorted_bam, output_vcf

     output_bam.unlink(missing_ok=True)
     log("BAM and VCF generation complete.", log_func)
+    return sorted_bam, output_vcf

src/streamlit_app.py CHANGED Viewed

@@ -5,6 +5,213 @@ import subprocess
 from pathlib import Path
 import pipeline as pl
 st.set_page_config(
     page_title="In Silico Controls Generator",
@@ -57,6 +264,17 @@ with st.sidebar:
             placeholder="/data/references/hg38.fa",
             help="Must be an indexed FASTA (.fa + .fa.fai).",
         )
 # ── Main area ────────────────────────────────────────────────────────────────
@@ -83,6 +301,87 @@ with col_info:
     if not hg38_cached and ref_mode == "Use cached / download":
         st.warning("hg38 not cached. First run will download ~938 MB and may take 5–10 minutes.")
 st.divider()
 # ── Step 2: Run pipeline ──────────────────────────────────────────────────────
@@ -117,9 +416,25 @@ if run_btn:
             st.warning("No .fai index found. Attempting to index with samtools faidx...")
             subprocess.run(f"samtools faidx {fasta_path}", shell=True, capture_output=True)
     work_dir = Path(tempfile.mkdtemp(prefix="insilicocontrols_"))
     probes_bed = work_dir / "probes.bed"
-    probes_bed.write_bytes(uploaded_bed.read())
     log_expander = st.expander("Pipeline log", expanded=True)
     log_area = log_expander.empty()
@@ -209,6 +524,18 @@ if run_btn:
         igv_bed_path = Path(snvs_bed) if not isinstance(snvs_bed, Path) else snvs_bed
         fully_bed_path = Path(fully_bed) if not isinstance(fully_bed, Path) else fully_bed
         # Store paths only — never load large files into session_state memory
         st.session_state["results"] = {
             "stats": stats,
@@ -218,6 +545,7 @@ if run_btn:
             "vcf_path": str(vcf_path),
             "igv_bed_path": str(igv_bed_path),
             "fully_covered_bed_path": str(fully_bed_path),
         }
         st.session_state["log_lines"] = log_lines[:]
@@ -247,6 +575,10 @@ if "results" in st.session_state:
     m3.metric("Off-target Probes", f"{stats['probes_no_exons']:,}")
     m4.metric("Total SNVs Generated", f"{total_snvs:,}")
     st.header("4 · Download Outputs")
     dl1, dl2, dl3 = st.columns(3)
@@ -327,4 +659,4 @@ st.caption(
     "intronic positions. For unused probes, a variant is placed at the midpoint. "
     "Paired-end reads are generated at the target depth and VAF, then written to "
     "a sorted, indexed BAM alongside a matching VCF."
-)

 from pathlib import Path
 import pipeline as pl
+import json
+import shutil
+def render_igv(res):
+    work_dir_name = res.get("work_dir_name")
+    if not work_dir_name:
+        st.warning("No static files path found for IGV.js. Restart the pipeline to generate.")
+        return
+    igv_bed_path = Path(res["igv_bed_path"])
+    variants = []
+    if igv_bed_path.exists():
+        with open(igv_bed_path, "r") as f:
+            for line in f:
+                if line.startswith("#") or not line.strip():
+                    continue
+                parts = line.strip().split("\t")
+                if len(parts) >= 4:
+                    chrom = parts[0]
+                    start = int(parts[1])
+                    end = int(parts[2])
+                    label = parts[3]
+                    locus = f"{chrom}:{max(1, start-50)}-{end+50}"
+                    variants.append({
+                        "locus": locus,
+                        "name": label.split("_")[-1],
+                        "label": label,
+                        "pos_label": f"{chrom}:{start+1}"
+                    })
+    variants_json = json.dumps(variants)
+    bam_url = f"/app/static/{work_dir_name}/synthetic.sorted.bam"
+    bai_url = f"/app/static/{work_dir_name}/synthetic.sorted.bam.bai"
+    vcf_url = f"/app/static/{work_dir_name}/synthetic.vcf"
+    navigator_url = f"/app/static/{work_dir_name}/igv_variant_navigator.bed"
+    probes_url = f"/app/static/{work_dir_name}/fully_covered_exons.bed"
+    html_content = f"""
+    <!DOCTYPE html>
+    <html>
+    <head>
+        <meta charset="utf-8">
+        <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.15.4/css/all.min.css">
+        <style>
+            body {{
+                font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
+                margin: 0;
+                padding: 0;
+                display: flex;
+                height: 600px;
+                background-color: #ffffff;
+            }}
+            #sidebar {{
+                width: 250px;
+                border-right: 1px solid #e0e0e0;
+                display: flex;
+                flex-direction: column;
+                height: 100%;
+                background-color: #f8f9fa;
+            }}
+            #sidebar-header {{
+                padding: 10px;
+                background-color: #0e1117;
+                color: white;
+                font-weight: bold;
+                font-size: 14px;
+            }}
+            #variant-list {{
+                flex-grow: 1;
+                overflow-y: auto;
+                padding: 5px;
+            }}
+            .variant-item {{
+                padding: 8px 10px;
+                margin-bottom: 4px;
+                border-radius: 4px;
+                cursor: pointer;
+                border: 1px solid #e9ecef;
+                background-color: white;
+                font-size: 12px;
+                transition: background-color 0.2s;
+            }}
+            .variant-item:hover {{
+                background-color: #e9ecef;
+            }}
+            .variant-name {{
+                font-weight: bold;
+                color: #ff4b4b;
+            }}
+            .variant-pos {{
+                color: #6c757d;
+                margin-top: 2px;
+            }}
+            #igv-container {{
+                flex-grow: 1;
+                height: 100%;
+                overflow: hidden;
+            }}
+            #igv-div {{
+                height: 600px;
+                width: 100%;
+            }}
+        </style>
+    </head>
+    <body>
+        <div id="sidebar">
+            <div id="sidebar-header"><i class="fas fa-list"></i> Variant Navigator ({len(variants)})</div>
+            <div id="variant-list"></div>
+        </div>
+        <div id="igv-container">
+            <div id="igv-div"></div>
+        </div>
+        <script src="https://cdn.jsdelivr.net/npm/igv@2.15.5/dist/igv.min.js"></script>
+        <script>
+            var variants = {variants_json};
+            var listContainer = document.getElementById("variant-list");
+            variants.forEach(function(v, index) {{
+                var item = document.createElement("div");
+                item.className = "variant-item";
+                item.innerHTML = '<div class="variant-name">' + v.name.toUpperCase() + '</div>' +
+                                 '<div class="variant-pos">' + v.pos_label + '</div>';
+                item.onclick = function() {{
+                    if (window.igvBrowser) {{
+                        window.igvBrowser.search(v.locus);
+                    }}
+                }};
+                listContainer.appendChild(item);
+            }});
+            var options = {{
+                genome: "hg38",
+                locus: variants.length > 0 ? variants[0].locus : "chr1:1787315-1787437",
+                tracks: [
+                    {{
+                        name: "Reference",
+                        type: "sequence",
+                        order: 1
+                    }},
+                    {{
+                        name: "Probes BED",
+                        type: "annotation",
+                        format: "bed",
+                        url: "{probes_url}",
+                        indexed: false,
+                        order: 2,
+                        color: "blue"
+                    }},
+                    {{
+                        name: "Variant Navigator BED",
+                        type: "annotation",
+                        format: "bed",
+                        url: "{navigator_url}",
+                        indexed: false,
+                        order: 3,
+                        color: "red"
+                    }},
+                    {{
+                        name: "Synthetic VCF",
+                        type: "variant",
+                        format: "vcf",
+                        url: "{vcf_url}",
+                        indexed: false,
+                        order: 4
+                    }},
+                    {{
+                        name: "Synthetic BAM",
+                        type: "alignment",
+                        format: "bam",
+                        url: "{bam_url}",
+                        indexURL: "{bai_url}",
+                        order: 5,
+                        height: 300
+                    }}
+                ]
+            }};
+            var igvDiv = document.getElementById("igv-div");
+            igv.createBrowser(igvDiv, options)
+                .then(function (browser) {{
+                    window.igvBrowser = browser;
+                    console.log("IGV browser created successfully.");
+                }})
+                .catch(function(err) {{
+                    console.error("Error creating IGV browser:", err);
+                    document.getElementById("igv-div").innerHTML =
+                        "<div style='color:#721c24; background-color:#f8d7da; border:1px solid #f5c6cb; padding:20px; border-radius:4px; font-family:sans-serif; margin:20px;'>" +
+                        "<h3>❌ Error Loading IGV Browser</h3>" +
+                        "<p><b>Message:</b> " + err.toString() + "</p>" +
+                        "<p>This usually indicates static file serving is not enabled or files are not accessible.</p>" +
+                        "<p><b>Paths attempted:</b></p>" +
+                        "<ul>" +
+                        "<li>BAM: <code>" + options.tracks[4].url + "</code></li>" +
+                        "<li>VCF: <code>" + options.tracks[3].url + "</code></li>" +
+                        "<li>BED: <code>" + options.tracks[1].url + "</code></li>" +
+                        "</ul>" +
+                        "<p>Please verify that <code>enableStaticServing = true</code> is active and the Hugging Face Space has fully rebuilt.</p>" +
+                        "</div>";
+                }});
+        </script>
+    </body>
+    </html>
+    """
+    st.components.v1.html(html_content, height=620, scrolling=False)
 st.set_page_config(
     page_title="In Silico Controls Generator",
             placeholder="/data/references/hg38.fa",
             help="Must be an indexed FASTA (.fa + .fa.fai).",
         )
+    st.divider()
+    st.subheader("🛠️ Debug Info")
+    st.caption("Helpful diagnostics for troubleshooting deployment status.")
+    st.write("Streamlit Version:", st.__version__)
+    st.write("File Uploaded:", uploaded_bed is not None)
+    if uploaded_bed:
+        st.write("Filename:", uploaded_bed.name)
+    st.write("probes_df in state:", "probes_df" in st.session_state)
+    if "probes_df" in st.session_state:
+        st.write("probes_df count:", len(st.session_state["probes_df"]))
 # ── Main area ────────────────────────────────────────────────────────────────
     if not hg38_cached and ref_mode == "Use cached / download":
         st.warning("hg38 not cached. First run will download ~938 MB and may take 5–10 minutes.")
+# ── Step 1.5: Customize Probes ────────────────────────────────────────────────
+if uploaded_bed:
+    if "probes_df" not in st.session_state or st.session_state.get("uploaded_file_name") != uploaded_bed.name:
+        import pandas as pd
+        import io
+        try:
+            uploaded_bed.seek(0)
+            content = uploaded_bed.read().decode("utf-8", errors="ignore")
+            # Strip comments and headers
+            lines = [line for line in content.splitlines() if line.strip() and not line.startswith("#") and not line.startswith("track")]
+            if lines:
+                df = pd.read_csv(io.StringIO("\n".join(lines)), sep="\t", header=None)
+                cols = ["chrom", "start", "end"]
+                if len(df.columns) > 3:
+                    cols += [f"col_{i}" for i in range(3, len(df.columns))]
+                df.columns = cols[:len(df.columns)]
+                df.insert(0, "Select", True)
+                st.session_state["probes_df"] = df
+                st.session_state["uploaded_file_name"] = uploaded_bed.name
+                st.session_state.pop("sample_seed", None)
+                st.session_state.pop("prev_frac", None)
+            else:
+                st.error("Uploaded BED file appears to be empty or contains only comments.")
+        except Exception as e:
+            st.error(f"Error parsing BED file: {e}")
+    if "probes_df" in st.session_state:
+        df = st.session_state["probes_df"]
+        st.header("1.5 · Customize Probes")
+        st.caption(f"Loaded {len(df):,} probes from {uploaded_bed.name}. Customize which regions will be processed below.")
+        col_mode, col_rand = st.columns([1, 1])
+        with col_mode:
+            subset_mode = st.radio(
+                "Selection mode",
+                options=["All Probes", "Manual Selection (below)", "Random Sampling"],
+                index=0,
+                help="Choose whether to run all probes, manually check/uncheck probes in the list, or select a random fraction of the probes."
+            )
+        with col_rand:
+            if subset_mode == "Random Sampling":
+                sample_frac = st.slider("Fraction of probes to keep", min_value=0.01, max_value=1.00, value=0.10, step=0.01)
+                resample_btn = st.button("🎲 Resample")
+                if "sample_seed" not in st.session_state or resample_btn or st.session_state.get("prev_frac") != sample_frac:
+                    import random
+                    st.session_state["sample_seed"] = random.randint(0, 100000)
+                    st.session_state["prev_frac"] = sample_frac
+                sampled_df = df.sample(frac=sample_frac, random_state=st.session_state["sample_seed"])
+                df["Select"] = df.index.isin(sampled_df.index)
+            elif subset_mode == "All Probes":
+                df["Select"] = True
+        # Render table editor
+        st.markdown("#### 📋 Probes List")
+        st.caption("Double-click a cell to search, or check/uncheck boxes to filter targets.")
+        edited_df = st.data_editor(
+            df,
+            use_container_width=True,
+            hide_index=True,
+            disabled=[col for col in df.columns if col != "Select"],
+            column_config={
+                "Select": st.column_config.CheckboxColumn(
+                    "Select",
+                    help="Uncheck to exclude this region from variant generation",
+                    default=True
+                )
+            }
+        )
+        st.session_state["probes_df"] = edited_df
+        total_selected = len(edited_df[edited_df["Select"] == True])
+        st.info(f"Selected {total_selected:,} of {len(df):,} probes ({total_selected/len(df)*100:.1f}%) for variant generation.")
 st.divider()
 # ── Step 2: Run pipeline ──────────────────────────────────────────────────────
             st.warning("No .fai index found. Attempting to index with samtools faidx...")
             subprocess.run(f"samtools faidx {fasta_path}", shell=True, capture_output=True)
+    # Filter for selected probes
+    if "probes_df" in st.session_state:
+        df = st.session_state["probes_df"]
+        selected_df = df[df["Select"] == True]
+    else:
+        st.error("No probe data found in session state.")
+        st.stop()
+    if len(selected_df) == 0:
+        st.error("No probes selected! Please select at least one probe in Step 1.5.")
+        st.stop()
+    # Convert back to BED format (tab-separated, without the 'Select' column)
+    bed_cols = [col for col in selected_df.columns if col != "Select"]
+    bed_text = selected_df[bed_cols].to_csv(sep="\t", header=False, index=False)
     work_dir = Path(tempfile.mkdtemp(prefix="insilicocontrols_"))
     probes_bed = work_dir / "probes.bed"
+    probes_bed.write_text(bed_text)
     log_expander = st.expander("Pipeline log", expanded=True)
     log_area = log_expander.empty()
         igv_bed_path = Path(snvs_bed) if not isinstance(snvs_bed, Path) else snvs_bed
         fully_bed_path = Path(fully_bed) if not isinstance(fully_bed, Path) else fully_bed
+        # Copy to static directory for IGV.js visualization
+        work_dir_name = work_dir.name
+        static_dest = Path("static") / work_dir_name
+        static_dest.mkdir(parents=True, exist_ok=True)
+        shutil.copy(sorted_bam, static_dest / "synthetic.sorted.bam")
+        if bai_path.exists():
+            shutil.copy(bai_path, static_dest / "synthetic.sorted.bam.bai")
+        shutil.copy(vcf_path, static_dest / "synthetic.vcf")
+        shutil.copy(igv_bed_path, static_dest / "igv_variant_navigator.bed")
+        shutil.copy(fully_bed_path, static_dest / "fully_covered_exons.bed")
         # Store paths only — never load large files into session_state memory
         st.session_state["results"] = {
             "stats": stats,
             "vcf_path": str(vcf_path),
             "igv_bed_path": str(igv_bed_path),
             "fully_covered_bed_path": str(fully_bed_path),
+            "work_dir_name": work_dir_name,
         }
         st.session_state["log_lines"] = log_lines[:]
     m3.metric("Off-target Probes", f"{stats['probes_no_exons']:,}")
     m4.metric("Total SNVs Generated", f"{total_snvs:,}")
+    st.header("🔍 Interactive Variant Browser")
+    st.caption("Inspect the generated synthetic alignments and mutations directly in the browser. Click on a variant in the navigator panel to jump to its locus.")
+    render_igv(res)
     st.header("4 · Download Outputs")
     dl1, dl2, dl3 = st.columns(3)
     "intronic positions. For unused probes, a variant is placed at the midpoint. "
     "Paired-end reads are generated at the target depth and VAF, then written to "
     "a sorted, indexed BAM alongside a matching VCF."
+)