31:T7e1,md`# Drug Pair Case Study: Toxicogenomics of Rosiglitazone and Troglitazone --- ## **Purpose** The tool facilitates a comparative analysis between **Rosiglitazone** and **Troglitazone**, two structurally related thiazolidinediones with divergent safety profiles. While both drugs act as PPAR-${tex`\gamma`} agonists, Troglitazone was withdrawn from the market due to severe **hepatotoxicity**, whereas Rosiglitazone remains in use with specific cardiovascular considerations. This notebook provides the functional "ground truth" for the framework by visualizing pathway-level perturbations captured via **RNA-sequencing** in HepaRG cells. ## Description of the Experiments 1) An **MTT experiment** with both drugs on HepaRG cells was performed and their respective IC20 and IC50 levels were determined: | Drug | IC20 (µM) | IC50 (µM) | | :--- | :--- | :--- | | Rosiglitazone | 216 | 338 | | Troglitazone | 42 | 56 | 2) **RNA-sequencing** was done on the HepaRG cells that were exposed to Rosiglitazone and Troglitazone at their respective IC20 and IC50 doses (i.e. equitoxic level for both drugs). The raw readcounts were processed and fold changes for each gene were calculated. ## **Interactive Controls** The notebook allows researchers to interactively explore the RNA-sequencing data: * **Experimental Conditions:** Explore responses at different equitoxic levels (**IC20** or **IC50**) and across various exposure durations ranging from **24 to 144 hours**. * **Criteria for deregulated genes:** Use the **log2 FC** and **Adjusted p-value** sliders to dynamically define and highlight significantly deregulated genes. ## **Data Infrastructure** This tool is built on the **Observable** platform and integrates directly with the **EdelweissData** infrastructure. It programmatically fetches differential expression data via the **Edelweiss Connect API**, ensuring that the analysis is grounded in **FAIR-compliant** (Findable, Accessible, Interoperable, and Reusable) data management practices. --- `32:T895,{ const totalCountChart = vl.markBar() .data(fold_change_data) .title("All DEGs") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().count().title("Number of DEGs"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) const PositiveFCChart = vl.markBar() .data(fold_change_data.filter(item => item["log2FoldChange"] > 0)) .title("Upregulated") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().count().title("Number of DEGs"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) const NegativeFCChart = vl.markBar() .data(fold_change_data.filter(item => item["log2FoldChange"] < 0)) .title("Downregulated") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().count().title("Number of DEGs"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) return vl.hconcat(totalCountChart, PositiveFCChart, NegativeFCChart) .resolve({ scale: { y: "shared" } // This links the Y-axis minimum and maximum boundaries across all panels }) .render() }33:T8cb,{ const totalCountChart = vl.markBar() .data(fold_change_data) .title("All DEGs") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().mean("log2FoldChange").title("Mean log2 Fold Change"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) const PositiveFCChart = vl.markBar() .data(fold_change_data.filter(item => item["log2FoldChange"] > 0)) .title("Upregulated") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().mean("log2FoldChange").title("Mean log2 Fold Change"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) const NegativeFCChart = vl.markBar() .data(fold_change_data.filter(item => item["log2FoldChange"] < 0)) .title("Downregulated") .encode( // X-axis shows the Drug names side-by-side in each subplot vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), // Y-axis counts thnumber of records (genes) vl.y().mean("log2FoldChange").title("Mean log2 Fold Change"), // Color bars by Drug for visual clarity vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) // Width of each individual small bar chart .height(150) return vl.hconcat(totalCountChart, PositiveFCChart, NegativeFCChart) .resolve({ scale: { y: "shared" } // This links the Y-axis minimum and maximum boundaries across all panels }) .render() }34:T76c,--- ***Findings:*** * *Starting with the lower dose, at IC20 and 24 h there is almost 600 DEGs (|logFC|>2, padj<0.05) with Rosiglitazone and only 7 with Troglitazone. The trend is similar for 48 h and 72 h. There is over 400 downregulated genes and less than 200 upregulated.* * *At 144 h the total number of DEGs decreases significantly. It was separately shown that even the blank sample at 144 h deteriorates substantially from the 24 h samples. So the main change at 144 h is due to these background processes. It is also possible that drugs are already metabolized and removed by the 144 h timepoint.* * *At IC50 the number of DEGs with Troglitazone jumps to 2,000 and is significantly higher than with Rosiglitazone (1,500). This is especially significant because an increase from IC20 to IC50 means only 14 µM increase in the dose of Troglitazone, while for Rosiglitazone if means 122 µM.* * *Another striking difference emerges between Rosiglitazone and Troglitazone at IC50: the magnitude of fold change with Rosiglitazone is close to 0 with roughly balanced numbers of up- and down-regulated genes (~900 and ~600). This could be understood as a balanced response of hepatic cells that are trying to accommodate to the stress. However with Troglitazone the average fold change is strongly negative and practically all DEGs are downregulated! This illustrates very clearly that mechanisms of toxicity will be completely different with the two drugs.* * *This observation also suggests that with Troglitazone at IC20 we would likely observe qualitatively the emerging effects, but perhaps the criteria for DEGs was too strict. Indeed, if we lower the criteria (e.g |logFC|>1 and padj<0.5) the pattern starts to appear. This will be imporant later for pathway enrichment analysis, because it justifies a different choice for enrichment algorithm in the case of Troglitazone at IC20.*35:Tbd5,--- ## Pathway Enrichment Analysis **Purpose:** we connected gene expression profiles with the pathways using two pathway enrichment algorithms and two gene sets (KEGG and WikiPathways). **Enrichment Algorithm:** Both methods answer the same core question: "Which biological pathways are significantly active or altered in my experiment?" However, they look at your data through completely different lenses. * **Over-Representation Analysis (ORA)** looks only at a small, curated list of "statistically significant" genes (a cut-off approach). The Question ORA Answers: *"Is the percentage of pathway genes in my list of significant genes higher than what I would expect by pure random chance?"*. The ORA was applied to a list of DEGs that were selected as |logFC| > 1 and padj < 0.05. * **Gene Set Enrichment Analysis (GSEA) Preranked** looks at every single gene you measured, sorted from best to worst (a global rank approach). The Question GSEA Answers: *"Do the genes in this pathway cluster coordinately at the extremes (the very top or very bottom) of my ranked list, or are they scattered randomly throughout?"*. This analysis was done to cover the case of Troglitazone at IC20 with a very small number of actual DEGs. **Pathway Activation:** was estimated using the following Z-Score formula: ${tex`Z = \frac{N_{up} - N_{down}}{\sqrt{N}}`} where: * ${tex`N_{up}`} is the number of upregulated genes in the pathway. * ${tex`N_{down}`} is the number of downregulated genes in the pathway. * ${tex`N`} is the total number of genes in the pathway.` Note that Z-score was calculated only for the pathways obtained through ORA analysis, while GSEA preranked algorithm returned Normalized Enrichment Score (NES) which is also a measure of pathway activation. **Pathway Selection:** To isolate the most robust and statistically sound biological signals, use the sliders below to dynamically restrict the pathways: * **Max adjusted p-value:** Controls the stringency of your False Discovery Rate (FDR). Lower values ensure changes are not due to experimental noise. * **Min Overlapping Genes:** minimum number of DEGs that are present in the pathway. * **Combined score:** a combination of significance (padj) and magnitude (Z-score). * **Gene set:** pick a single gene set for enrichment analysis (KEGG or WikiPathways). **Visualization:** * **Count of Affected Pathways:** count of selected pathways and separately for activated (Z-score > 0) and inactivated (Z-score < 0) pathways. * **Magnitude of Pathway Activation:** average Z-score of all selected pathways and separately magnitude of activated (Z-score > 0) and inactivated pathways (Z-score < 0). * **Heatmap of Pathway Activation:** individual pathways (in rows) for each drug (in columns) with their respective Z-score (colour). There is one heatmap for each enrichment algorithm: the ORA approach the GSEA preranked approach. Note that with the pathways from ORA analysis we used Z-score and for pathways from GSEA preranked analysis we used NES. ---36:T621,{ const totalCountChart = vl.markBar() .data(filtered_ora_data) .title("All pathways") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().count().title("Number of pathways"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); const PositiveZChart = vl.markBar() .data(filtered_ora_data.filter(item => item['Z-score'] > 0)) .title("Activated (Z-score > 0)") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().count().title("Number of pathways"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); const NegativeZChart = vl.markBar() .data(filtered_ora_data.filter(item => item['Z-score'] < 0)) .title("Inactivated (Z-score < 0)") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().count().title("Number of pathways"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip([vl.y().count(), "Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); return vl.hconcat(totalCountChart, PositiveZChart, NegativeZChart) .resolve({ scale: { y: "shared" } // This links the Y-axis minimum and maximum boundaries across all panels }) .render() }37:T600,{ const totalCountChart = vl.markBar() .data(filtered_ora_data) .title("All pathways") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().mean("Z-score").title("Average Z-score"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip(["Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); const PositiveZChart = vl.markBar() .data(filtered_ora_data.filter(item => item['Z-score'] > 0)) .title("Activated (Z-score > 0)") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().mean("Z-score").title("Average Z-score"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip(["Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); const NegativeZChart = vl.markBar() .data(filtered_ora_data.filter(item => item['Z-score'] < 0)) .title("Inactivated (Z-score < 0)") .encode( vl.x().fieldN("Drug").axis({ labelAngle: 0 }).title(null), vl.y().mean("Z-score").title("Average Z-score"), vl.color() .fieldN("Drug") .scale({ range: ["steelblue", "#ff7f0e"] }), // Blue and Orange vl.tooltip(["Drug", "Dose", "Exposure_time_hr"]) ) .width(120) .height(150); return vl.hconcat(totalCountChart, PositiveZChart, NegativeZChart) .resolve({ scale: { y: "shared" } // This links the Y-axis minimum and maximum boundaries across all panels }) .render() }38:Tc49,--- ***Findings for Rosiglitazone:*** * *At IC20, 24 h we identified 7 KEGG pathways for Rosiglitazone (padj < 0.05, overlapping genes >= 3, combined score > 50). Most of these pathways were somewhat inactivated (average Z-score around -1.5).* * *The most inactivated KEGG pathway at IC20 is "Complement and coagulation cascades" which reflects the targeted anti-inflammatory and protective threshold of Rosiglitazone. By suppressing these inflammatory cascades, the drug avoids the severe, acute hepatotoxic profile seen in earlier generation analogs like Troglitazone, highlighting a successful dampening of innate immune stress.* * *The only activated KEGG pathway at IC20 is "ECM-receptor interaction", which captures the structural tissue-remodeling events driven by PPAR-${tex`\gamma`} stimulation. This path is crucial for adipocyte differentiation, but its manifestation in vascular tissue explains the compound's clinical propensity for causing fluid retention and edema (these were clinically observed side effects of Rosiglitazone).* * *At IC50 the total number of affected pathways is only 3 and they are all activated: Ribosome, Mitophagy and Coronavirus disease.* * *Activation of Mitophagy is very telling: we know from other studies (such as reporter assays) that Rosiglitazone causes mitochondrial stress. But the cells successfully activate Mitophagy to quarantine and digest the damaged mitochondria before they can spill reactive oxygen species (ROS) or trigger necrotic cell death. The cell is actively cleaning its house.* * *An activated Ribosome pathway (Z-score > 3) means the HepaRG cells are aggressively synthesizing new proteins. They are actively translating adaptive proteins to survive the IC50 stress, rather than passively dying.* * *In KEGG, the "Coronavirus disease" pathway is heavily populated by genes encoding ribosomal proteins, translation initiation factors, and broad-spectrum inflammatory/stress-responsive cytokines. At IC50, the activation of this pathway indicates a generic, controlled Integrated Stress Response. The cell has triggered its alarm system, synthesized ribosomal proteins (hence the overlap with the Ribosome pathway), and is actively managing the stress.* ***Findings for Troglitazone:*** * *With the ORA approach we don't find any enriched pathway at IC20, 24 h, because there is only 29 DEGs at this condition.* * *To capture the effect of numerous small changes in gene expression we employed GSEA preranked approach instead of the cut-off-based ORA. With this approach we identified 7 enriched pathways among which all of them are deactivated. Majority of these pathways are related to metabolism (cholesterol, choline, glyoxylate, aminoacids).* * *At IC50, 24 h the number of enriched pathways is 29 and all of them are deactivated. Majority of deactivated pathways are related to various metabolisms. This is signalling very strongly that HepaRG cells are running out of energy, which correlates very well with the clinical outcomes of a catastrophic liver failure for Troglitazone.* --- **Extra plot:** Common deregulated genes in a heatmap (both drugs, time profiles)39:[{"id":"9b51690a139c79b0a05079a6fb7207ce3196631abdebe6ecd8ed2f6eae0f9ec1f5001d912463c79a86c8c64852673afa0d779a6198975e1ba520a8db90c3620b","href":"https://static.observableusercontent.com/files/9b51690a139c79b0a05079a6fb7207ce3196631abdebe6ecd8ed2f6eae0f9ec1f5001d912463c79a86c8c64852673afa0d779a6198975e1ba520a8db90c3620b","name":"image.png","type":"image/png","lastModified":1779632582264.619,"size":10329}] 2f:{"title":"Drug Pair Case Study: Toxicogenomics of Rosiglitazone and Troglitazone","cells":[{"id":4,"value":"$31","pinned":false,"hidden":false,"mode":"ojs"},{"id":604,"value":"## Experimental conditions","pinned":false,"hidden":false,"mode":"md"},{"id":353,"value":"viewof selected_dose = Select(unique_doses, {label: \"Dose\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":360,"value":"viewof selected_timepoint = Select(unique_timepoints, {label: \"Timepoint\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":212,"value":"## Criteria for deregulated genes","pinned":false,"hidden":false,"mode":"md"},{"id":115,"value":"viewof threshold_logfc = Inputs.range([0, 10], {label: \"Min. abs. log2 FC\", step: 0.5})","pinned":false,"hidden":false,"mode":"ojs"},{"id":117,"value":"viewof threshold_padj = Inputs.range([0, 1], {label: \"Max. adj. p-value\", step: 0.01})","pinned":false,"hidden":false,"mode":"ojs"},{"id":214,"value":"---\n\n## Gene Expression Profiles\n\n**Purpose:** comparison of differential gene expression between Rosiglitazone and Troglitazone.\n\n\n**Visualizations:**\n* **Differentially Expressed Gene (DEG) Count:** count of all DEG and separately count of DEG with positive and negative fold change.\n* **Magnitude of the Fold Change:** average fold change of all DEG and separately for DEG with positive and negative fold changes. This illustrates the magnitude of the fold change in each direction.\n* **Volcano plot:** significance of the fold change vs. the magnitude of the fold change. Only genes surpassing the thresholds are displayed.\n* **Heatmap of the Common DEGs:** magnitude and direction of the fold change for both drugs.\n\n---","pinned":false,"hidden":false,"mode":"md"},{"id":627,"value":"**Count of Differentially Expressed Genes (DEG)**","pinned":false,"hidden":false,"mode":"md"},{"id":111,"value":"$32","pinned":false,"hidden":false,"mode":"ojs"},{"id":629,"value":"**Magnitude of the Fold Change**","pinned":false,"hidden":false,"mode":"md"},{"id":577,"value":"$33","pinned":false,"hidden":false,"mode":"ojs"},{"id":633,"value":"**Volcano plot (significance vs. magnitude)**","pinned":false,"hidden":false,"mode":"md"},{"id":76,"value":"vl.markPoint({ filled: true, opacity: 0.6 })\n .data(fold_change_data)\n .transform(\n vl.calculate(\"-log(datum.padj) / log(10)\").as(\"negLogPadj\")\n )\n .encode(\n vl.x().fieldQ(\"log2FoldChange\").title(\"log2 Fold Change\"),\n vl.y().fieldQ(\"negLogPadj\").scale({ reverse: false }).title(\"-log10 p-adj\"), \n vl.column().fieldN(\"Drug\"),\n vl.tooltip([\"SYMBOL\", \"log2FoldChange\", \"negLogPadj\", \"padj\"])\n )\n .width(300)\n .height(300)\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":391,"value":"**Heatmap of the Common DEGs**","pinned":false,"hidden":false,"mode":"md"},{"id":782,"value":"viewof heatmap_height = Inputs.range([0, 10000], {label: \"Total height (adjust for better readability)\", step: 50})","pinned":false,"hidden":false,"mode":"ojs"},{"id":394,"value":"vl.markRect()\n .data(common_genes_dataset)\n .encode(\n vl.x().fieldN(\"Drug\").title(\"Drug\").axis({ labelAngle: 0 }),\n vl.y().fieldN(\"SYMBOL\").title(\"Gene Symbol\"),\n vl.color()\n .fieldQ(\"log2FoldChange\") \n .scale({ scheme: \"redblue\", midnum: 0, domainMid: 0, reverse: true }),\n vl.tooltip([\n vl.fieldN(\"Drug\"),\n vl.fieldN(\"SYMBOL\"),\n vl.fieldQ(\"log2FoldChange\"),\n vl.fieldQ(\"padj\")\n ])\n )\n .width(150)\n .height(heatmap_height)\n .config({ view: { stroke: null }, axis: { grid: false } })\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":372,"value":"**List of DEGs**","pinned":false,"hidden":false,"mode":"md"},{"id":168,"value":"viewof selected_drug = Select(unique_drugs)","pinned":false,"hidden":false,"mode":"ojs"},{"id":164,"value":"Table(fold_change_data.filter(d => \n d.padj !== null &&\n d.Drug == selected_drug),\n {\n columns: [\"SYMBOL\", \"log2FoldChange\", \"padj\"]\n }\n )","pinned":false,"hidden":false,"mode":"ojs"},{"id":640,"value":"$34","pinned":false,"hidden":false,"mode":"md"},{"id":526,"value":"$35","pinned":false,"hidden":false,"mode":"md"},{"id":712,"value":"**Pathway Selection**","pinned":false,"hidden":false,"mode":"md"},{"id":470,"value":"viewof threshold_padj_pathways = Inputs.range([0, 0.5], {label: \"Adjusted p-value threshold\", step: 0.01})","pinned":false,"hidden":false,"mode":"ojs"},{"id":474,"value":"viewof threshold_overlapping_genes = Inputs.range([0, 10], {label: \"Number of genes overlapping with the pathways\", step: 1})","pinned":false,"hidden":false,"mode":"ojs"},{"id":484,"value":"viewof threshold_combined_score = Inputs.range([0, 100], {label: \"Combined score\", step: 1})","pinned":false,"hidden":false,"mode":"ojs"},{"id":508,"value":"viewof selected_geneset = Select([\"KEGG_2021_Human\", \"WikiPathways_2019_Human\"], {label: \"Gene set\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":804,"value":"viewof selected_fold_change_direction = Select([\"All\", \"Down\", \"Up\"], {label: \"Fold change direction\"})","pinned":false,"hidden":false,"mode":"ojs"},{"id":500,"value":"---\n\n**Count of Affected Pathways**","pinned":false,"hidden":false,"mode":"md"},{"id":454,"value":"$36","pinned":false,"hidden":false,"mode":"ojs"},{"id":705,"value":"**Magnitude of Pathway Activation**","pinned":false,"hidden":false,"mode":"md"},{"id":537,"value":"$37","pinned":false,"hidden":false,"mode":"ojs"},{"id":718,"value":"**Heatmap of Pathway Activation (from ORA approach)**","pinned":false,"hidden":false,"mode":"md"},{"id":777,"value":"viewof row_height1 = Inputs.range([0, 6], {label: \"Row height (adjust for better readability)\", step: 0.1})","pinned":false,"hidden":false,"mode":"ojs"},{"id":463,"value":"vl.markRect()\n .data(filtered_ora_data)\n .encode(\n vl.x().fieldN(\"Drug\").title(\"Drug\").axis({ labelAngle: 0 }),\n vl.y().fieldN(\"Term\").title(\"Pathway\").axis({maxExtent: 350, labelLimit: 340}),\n vl.color()\n .fieldQ(\"Z-score\") \n .scale({ scheme: \"redblue\", midnum: 0, domainMid: 0, reverse: true }),\n vl.tooltip([\n vl.fieldN(\"Drug\"),\n vl.fieldN(\"Term\"),\n vl.fieldQ(\"Z-score\"),\n vl.fieldQ(\"Adjusted P-value\")\n ])\n )\n .width(150)\n .height(ora_data.length * row_height1)\n .config({ view: { stroke: null }, axis: { grid: false } })\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":524,"value":"**Heatmap of Pathway Activation (from GSEA prerank approach)**","pinned":false,"hidden":false,"mode":"md"},{"id":770,"value":"viewof row_height2 = Inputs.range([0, 12], {label: \"Row height (adjust for better readability)\", step: 0.1})","pinned":false,"hidden":false,"mode":"ojs"},{"id":519,"value":"vl.markRect()\n .data(filtered_prerank_data)\n .encode(\n vl.x().fieldN(\"Drug\").title(\"Drug\").axis({ labelAngle: 0 }),\n vl.y().fieldN(\"Term\").title(\"Pathway\").axis({maxExtent: 350, labelLimit: 340}),\n vl.color()\n .fieldQ(\"NES\") \n .scale({ scheme: \"redblue\", midnum: 0, domainMid: 0, reverse: true }),\n vl.tooltip([\n vl.fieldN(\"Drug\"),\n vl.fieldN(\"Term\"),\n vl.fieldQ(\"NES\"),\n vl.fieldQ(\"FDR q-val\")\n ])\n )\n .width(150)\n .height(ora_data.length * row_height2)\n .config({ view: { stroke: null }, axis: { grid: false } })\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":441,"value":"$38","pinned":false,"hidden":false,"mode":"md"},{"id":429,"value":"vl.data(common_genes_dataset)\n // 1. Tell the top-level chart to split into columns by Drug\n .facet(\n vl.column().fieldN(\"Drug\").title(\"Drug Treatment\")\n )\n // 2. Define the nested chart spec, explicitly declaring markRect() inside it\n .spec(\n vl.markRect().encode(\n vl.x().fieldO(\"Exposure_time_hr\").title(\"Exposure Time (hours)\"),\n vl.y().fieldN(\"SYMBOL\").title(\"Gene Symbol\"),\n vl.color()\n .fieldQ(\"log2FoldChange\") \n .scale({ scheme: \"redblue\", midnum: 0, reverse: true }),\n vl.tooltip([\n vl.fieldN(\"Drug\"),\n vl.fieldN(\"Exposure_time_hr\"),\n vl.fieldN(\"SYMBOL\"),\n vl.fieldQ(\"log2FoldChange\"),\n vl.fieldQ(\"padj\")\n ])\n )\n .width(150)\n .height(300)\n )\n .config({ view: { stroke: null }, axis: { grid: false } })\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":209,"value":"---\n\n## Notebook Code\n\n","pinned":false,"hidden":false,"mode":"md"},{"id":504,"value":"filtered_prerank_data = prerank_data.filter(item => {\n // 1. Keep rows where padj is LESS than or equal to the threshold\n const passesPadj = item[\"FDR q-val\"] <= threshold_padj_pathways;\n const passesGeneset = item[\"Geneset\"] == selected_geneset;\n \n // Both conditions must be true to keep the row\n return passesPadj && passesGeneset;\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":480,"value":"filtered_ora_data = ora_data_2.filter(item => {\n // 1. Keep rows where padj is LESS than or equal to the threshold\n const passesPadj = item[\"Adjusted P-value\"] <= threshold_padj_pathways;\n const passesCombinedScore = item[\"Combined Score\"] >= threshold_combined_score;\n const passesGeneset = item[\"Gene_set\"] == selected_geneset;\n // 2. Keep rows where overlapping_genes is GREATER than or equal to the threshold\n const passesOverlap = item.overlapping_genes >= threshold_overlapping_genes;\n \n // Both conditions must be true to keep the row\n return passesPadj && passesOverlap && passesCombinedScore && passesGeneset;\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":477,"value":"ora_data_2 = ora_data.map(item => {\n // Grab the first number before the slash\n const [overlappingCount] = item[\"Overlap\"].split('/');\n \n // Return a new object with the new numerical field added\n return {\n ...item,\n overlapping_genes: parseInt(overlappingCount, 10)\n };\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":60,"value":"unique_doses = [\"IC20\", \"IC50\"]","pinned":false,"hidden":false,"mode":"ojs"},{"id":58,"value":"unique_drugs = [\"Rosiglitazone\", \"Troglitazone\"]","pinned":false,"hidden":false,"mode":"ojs"},{"id":56,"value":"unique_timepoints = [24, 48, 72, 144]","pinned":false,"hidden":false,"mode":"ojs"},{"id":377,"value":"rosGenes = new Set(fold_change_data.filter(item => item.Drug === 'Rosiglitazone').map(item => item.SYMBOL));","pinned":false,"hidden":false,"mode":"ojs"},{"id":381,"value":"troGenes = new Set(fold_change_data.filter(item => item.Drug === 'Troglitazone').map(item => item.SYMBOL));","pinned":false,"hidden":false,"mode":"ojs"},{"id":375,"value":"commonGenes = new Set([...rosGenes].filter(gene => troGenes.has(gene)))","pinned":false,"hidden":false,"mode":"ojs"},{"id":387,"value":"fold_change_data_common_genes = fold_change_data.filter(item => commonGenes.has(item.SYMBOL));","pinned":false,"hidden":false,"mode":"ojs"},{"id":232,"value":"prerank_data = fetch_dataset_to_array(prerank_url, prerank_query)","pinned":false,"hidden":false,"mode":"ojs"},{"id":336,"value":"prerank_query = {\n return {\n method: 'POST',\n body: JSON.stringify({\n condition: {\n \"and\": [\n {\"exactSearch\": [{ \"column\": [\"Dose\"] }, selected_dose]},\n {\"eq\": [{ \"column\": [\"Exposure_time_hr\"] }, selected_timepoint]},\n {\"le\": [{ \"column\": [\"FDR q-val\"] }, threshold_padj_pathways]},\n ]\n },\n }),\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":312,"value":"ora_query = {\n return {\n method: 'POST',\n body: JSON.stringify({\n condition: {\n \"and\": [\n {\"exactSearch\": [{ \"column\": [\"Dose\"] }, selected_dose]},\n {\"exactSearch\": [{ \"column\": [\"Direction\"] }, selected_fold_change_direction]},\n {\"eq\": [{ \"column\": [\"Exposure_time_hr\"] }, selected_timepoint]},\n {\"le\": [{ \"column\": [\"Adjusted P-value\"] }, threshold_padj_pathways]},\n ]\n },\n }),\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":399,"value":"common_genes_dataset = fetch_dataset_to_array(fold_change_dataset_url, common_genes_query)","pinned":false,"hidden":false,"mode":"ojs"},{"id":396,"value":"common_genes_query = {\n return {\n method: 'POST',\n body: JSON.stringify({\n condition: {\n \"and\": [\n {\"exactSearch\": [{ \"column\": [\"Dose\"] }, selected_dose]},\n {\"containedIn\": [{ \"column\": [\"SYMBOL\"] }, Array.from(commonGenes)]},\n ]\n },\n }),\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":236,"value":"fold_change_query = {\n return {\n method: 'POST',\n body: JSON.stringify({\n condition: {\n \"and\": [\n {\"exactSearch\": [{ \"column\": [\"Dose\"] }, selected_dose]},\n {\"eq\": [{ \"column\": [\"Exposure_time_hr\"] }, selected_timepoint]},\n {\"le\": [{ \"column\": [\"padj\"] }, threshold_padj]},\n {\n \"or\": [\n {\"ge\": [{ \"column\": [\"log2FoldChange\"] }, threshold_logfc]},\n {\"le\": [{ \"column\": [\"log2FoldChange\"] }, -threshold_logfc]}\n ]\n }\n\n ]\n },\n }),\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":230,"value":"ora_data = fetch_dataset_to_array(ora_url, ora_query)","pinned":false,"hidden":false,"mode":"ojs"},{"id":6,"value":"fold_change_data = fetch_dataset_to_array(fold_change_dataset_url, fold_change_query) ","pinned":false,"hidden":false,"mode":"ojs"},{"id":7,"value":"function fetch_dataset_to_array(dataset_url, query_options) {\n return fetch(dataset_url, query_options).then(response_dataset => {\n if (!response_dataset.ok) throw new Error(response_dataset.status);\n return response_dataset.json();}).then(json => json.results.map(x => x.data));\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":225,"value":"ora_url = api_url + ora_dataset_id + \"/versions/latest/data\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":222,"value":"prerank_url = api_url + prerank_dataset_id + \"/versions/latest/data\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":49,"value":"fold_change_dataset_url = api_url + fold_change_dataset_id + \"/versions/latest/data\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":40,"value":"api_url = \"https://api.kit.edelweissconnect.com/datasets/\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":220,"value":"prerank_dataset_id = \"8031a6a0-c70b-4c01-a3b5-56cb37160faf\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":218,"value":"ora_dataset_id = \"a13e8d7f-de5d-4427-9be2-de78fa1be701\" //\"6c6d1326-7e2d-4cfd-a02b-314485f95868\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":25,"value":"fold_change_dataset_id = \"45d032c3-4cd5-4e7f-be25-ff5ece93bb28\"//\"a4f41901-d3d8-46ce-910d-1b927529fdae\" ","pinned":false,"hidden":false,"mode":"ojs"},{"id":21,"value":"import {Select, Range, Toggle, Search, Table, Text} from \"@observablehq/inputs\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":16,"value":"import {vl} from \"@vega/vega-lite-api\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":12,"value":"import {html} from \"@observablehq/htl\"","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W39","stdlib":"1","titling":"auto"}