From 7619607e809d627ea647f58486183e3d736411c8 Mon Sep 17 00:00:00 2001 From: Jobin Sunny Date: Mon, 9 Jun 2025 13:44:53 -0500 Subject: [PATCH 1/5] docs: add FAQ for expression data facet. Clean up some markdown and text --- .../3.pecan/1.overview/2.methods-and-data.md | 14 +++++----- content/3.pecan/2.data-facets/3.expression.md | 28 +++++++++++++------ 2 files changed, 26 insertions(+), 16 deletions(-) diff --git a/content/3.pecan/1.overview/2.methods-and-data.md b/content/3.pecan/1.overview/2.methods-and-data.md index 052c77a..aaddc0e 100644 --- a/content/3.pecan/1.overview/2.methods-and-data.md +++ b/content/3.pecan/1.overview/2.methods-and-data.md @@ -18,7 +18,7 @@ Variant data is sourced from published studies from St. Jude Children's Research A user can access associated studies by clicking `Associated Study` where applicable. ::callout{icon="i-heroicons-light-bulb" color="blue"} -**Tip** +**Tip** Gene lists are not curated. Access specific `Associated Study` links for subsets of data, such as the Medulloblastoma example [here](https://pecan.stjude.cloud/variants/oncoprint/BT%7CMB). :: @@ -45,7 +45,7 @@ The 20 most frequently observed genes are shown by default in each pathway; othe | CNV GAIN | Copy number gain. | ::callout{icon="i-heroicons-exclamation-circle" color="orange"} -**Warning** +**Warning** Silent mutations and untranslated region (UTR) variants are excluded in some prevalence views due to their lack of effect. Only pathogenic or likely pathogenic variants are included in assessing mutational prevalence for pan-cancer genes. CNV classes are not displayed in ProteinPaint. @@ -143,7 +143,7 @@ COSMIC SBS signatures (v3.3) were identified using SigProfilerExtractor (v1.1.20 Sample data from G4K, PCGP, Clinical Pilot, and Real-Time Clinical Genomics (RTCG) were used for mutational signature detection. ::callout{icon="i-heroicons-light-bulb" color="blue"} -**Tip** +**Tip** Example data from 790 tumor samples can be found in [Figure 5 of McLeod et al](https://cancerdiscovery.aacrjournals.org/content/11/5/1082.long). :: @@ -153,20 +153,20 @@ Example data from 790 tumor samples can be found in [Figure 5 of McLeod et al](h ### t-SNE Methods -[RAPID RNA-Seq workflow](https://university.stjude.cloud/docs/genomics-platform/workflow-guides/rapid-rnaseq/) was used to generate RNA-seq expression count data and visualized via ProteinPaint's scatter plot. +St. Jude Cloud's [Rapid RNA-Seq workflow](https://university.stjude.cloud/docs/genomics-platform/workflow-guides/rapid-rnaseq/) was used to generate RNA-seq expression counts data and visualized on PeCan's expression t-SNE plot. ### t-SNE Data Sample data comes from G4K, PCGP, Clinical Pilot, and Real-Time Clinical Genomics studies. ::callout{icon="i-heroicons-light-bulb" color="blue"} -**Tip** +**Tip** Example data from 1,574 RNA-Seq samples can be accessed [here](https://viz.stjude.cloud/st-jude-childrens-research-hospital/visualization/pediatric-blood-solid-and-brain-tumor-rna-seq-t-sne-plot-1574-samples~24). :: ### Gene Expression Methods -DESeq2's Median of Ratios^1 was applied to ~5500 tumor samples for gene expression analysis. +DESeq2's Median of Ratios was applied to ~5500 tumor samples for gene expression analysis. We advocate for this methodology over TPM/CPM due to its stability and resistance to outliers. #### Key Benefits @@ -192,6 +192,6 @@ Histological images are provided by the [COMET Blue Sky Initiative](https://www. A large subset of COMET slides are pending publication. ::callout{icon="i-heroicons-exclamation-circle" color="green"} -**Interested in collaborating?** +**Interested in collaborating?** Reach out to us at [support@stjude.cloud](mailto:support@stjude.cloud). :: diff --git a/content/3.pecan/2.data-facets/3.expression.md b/content/3.pecan/2.data-facets/3.expression.md index f757262..11c6dd7 100644 --- a/content/3.pecan/2.data-facets/3.expression.md +++ b/content/3.pecan/2.data-facets/3.expression.md @@ -12,9 +12,8 @@ This facet comprises three tabs, allowing users to explore the expression landsc ![tsne sample view](/img/pecan/data-facets/expression/expression-t-sne-sample-view.png) -**Figure 1: t-SNE for Blood, Brain, and Solid Samples.** +**Figure 1: t-SNE for Blood, Brain, and Solid Tumor Samples.** Mouse over data points to access metadata details for each sample. -Visualization powered by D3. ![gene violin plots](/img/pecan/data-facets/expression/expression-gene-violin-plots.png) @@ -23,10 +22,11 @@ Gene expression violin plots for each sample, filtered by the gene of interest. Visualization powered by Plotly. ::callout{icon="i-heroicons-document-text-solid" color="blue"} -**Note** +**Note** - All samples use the hg38 reference genome. - Full metadata can be accessed through our [manifest](https://platform.stjude.cloud/api/v1/manifest). +- Gene Expression values are log10 transformed to normalize outlier values while representing the expression data. :: --- @@ -45,7 +45,7 @@ Visualization powered by Plotly. ![tsne features overview](/img/pecan/data-facets/expression/expression-t-sne-features-overview.gif) ::callout{icon="i-heroicons-exclamation-circle" color="orange"} -**Warning** +**Warning** Filtering by the sunburst will auto-populate the Root and Subtype filters. These can be manually edited but will not update the sunburst. :: @@ -61,7 +61,7 @@ These can be manually edited but will not update the sunburst. | **Median Sort** | Sort the gene expression sandboxes by median expression across or within individual groups. | | **Outlier Toggle** | Toggle off data points to keep outliers intact for the cohort currently being filtered. | -For data normalization details, refer to our [Methods and Data](/pecan/overview/methods-and-data) page. +For data normalization details, refer to our [Methods and Data](/pecan/overview/methods-and-data#gene-expression-methods) page. ![violin plots](/img/pecan/data-facets/expression/expression-violin-plots.gif) @@ -71,7 +71,7 @@ For data normalization details, refer to our [Methods and Data](/pecan/overview/ Users can overlay gene expression on the t-SNE plot by selecting genes of interest. Count data is normalized using Median of Ratios (MoR). -More details can be found on the [Methods and Data](/pecan/overview/methods-and-data) page. +More details can be found on the [Methods and Data](/pecan/overview/methods-and-data#gene-expression-methods) page. ![gene expression toggle](/img/pecan/data-facets/expression/expression-gene-expression-toggle.gif) @@ -114,15 +114,25 @@ The data matrix displays all filtered data with sortable headers for easier expl | **Preservative** | Multi-select dropdown for sample preservative types. | ::callout{icon="i-heroicons-exclamation-circle" color="orange"} -**Warning** +**Warning** Some fields may have a "Not Available" option for samples where the data wasn't recorded (e.g., Race, Ethnicity, Sex). :: ::callout{icon="i-heroicons-light-bulb" color="blue"} -**Tip** +**Tip** For a subset of this data, refer to [Figure 4f of McLeod et al.](https://cancerdiscovery.aacrjournals.org/content/11/5/1082.long) :: --- -To see how the data was calculated and normalized, visit our [Methods and Data](/pecan/overview/methods-and-data) page. +## Frequently Asked Questions + +### Why was the expression data transformed using log10? + +During development, we observed that displaying raw Median of Ratios values resulted in outliers disproportionately dominating the visualizations, making it difficult to distinguish differences among most samples. +To address this, we evaluated both log2 and log10 transformations of the data on a selection of important representative genes. The log10 transformation offered a slightly more effective and visually interpretable presentation of the data. +As a result, we adopted the log10 transformation for all Median of Ratios values displayed in the interface. + +--- + +To see how the data was calculated and normalized, visit our [Methods and Data](/pecan/overview/methods-and-data#expression) page. From 5c42b938518f844d3130556b6e6a59a6a19b22a6 Mon Sep 17 00:00:00 2001 From: Jobin Sunny Date: Mon, 9 Jun 2025 13:58:44 -0500 Subject: [PATCH 2/5] docs: revert whitespace deletion in note --- content/3.pecan/2.data-facets/3.expression.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/3.pecan/2.data-facets/3.expression.md b/content/3.pecan/2.data-facets/3.expression.md index 11c6dd7..75f8d29 100644 --- a/content/3.pecan/2.data-facets/3.expression.md +++ b/content/3.pecan/2.data-facets/3.expression.md @@ -22,7 +22,7 @@ Gene expression violin plots for each sample, filtered by the gene of interest. Visualization powered by Plotly. ::callout{icon="i-heroicons-document-text-solid" color="blue"} -**Note** +**Note** - All samples use the hg38 reference genome. - Full metadata can be accessed through our [manifest](https://platform.stjude.cloud/api/v1/manifest). From 0646525ab36f3f42c56eea08235564f78d5299c7 Mon Sep 17 00:00:00 2001 From: Jobin Sunny Date: Mon, 9 Jun 2025 14:37:25 -0500 Subject: [PATCH 3/5] docs: address PR feedback --- content/3.pecan/1.overview/2.methods-and-data.md | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/content/3.pecan/1.overview/2.methods-and-data.md b/content/3.pecan/1.overview/2.methods-and-data.md index aaddc0e..57291b0 100644 --- a/content/3.pecan/1.overview/2.methods-and-data.md +++ b/content/3.pecan/1.overview/2.methods-and-data.md @@ -153,7 +153,7 @@ Example data from 790 tumor samples can be found in [Figure 5 of McLeod et al](h ### t-SNE Methods -St. Jude Cloud's [Rapid RNA-Seq workflow](https://university.stjude.cloud/docs/genomics-platform/workflow-guides/rapid-rnaseq/) was used to generate RNA-seq expression counts data and visualized on PeCan's expression t-SNE plot. +St. Jude Cloud's [RNA-Seq V2 WDL workflow](https://github.com/stjudecloud/workflows/blob/main/workflows/rnaseq/rnaseq-standard.wdl) was used to generate HTSeq counts data for each sample and visualized on PeCan's expression t-SNE plot. ### t-SNE Data @@ -166,7 +166,7 @@ Example data from 1,574 RNA-Seq samples can be accessed [here](https://viz.stjud ### Gene Expression Methods -DESeq2's Median of Ratios was applied to ~5500 tumor samples for gene expression analysis. +DESeq2's Median of Ratios1 was applied to ~5500 tumor samples for gene expression analysis. We advocate for this methodology over TPM/CPM due to its stability and resistance to outliers. #### Key Benefits @@ -175,7 +175,8 @@ We advocate for this methodology over TPM/CPM due to its stability and resistanc 2. Robust against highly expressed genes dominating total read counts. 3. More reliable differential expression results due to geometric mean normalization. 4. Consistent and comparable expression values across samples. -5. Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. [https://doi.org/10.1186/s13059-014-0550-8](https://doi.org/10.1186/s13059-014-0550-8) + +1 Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. [https://doi.org/10.1186/s13059-014-0550-8](https://doi.org/10.1186/s13059-014-0550-8) --- From 94936d4d6d521cc8ac654a03fb6ee98fe18c95f8 Mon Sep 17 00:00:00 2001 From: Jobin Sunny <38107318+jsunny23@users.noreply.github.com> Date: Tue, 17 Jun 2025 14:42:58 -0500 Subject: [PATCH 4/5] docs: update RNASeq V2 workflow link to v2.0.0 tag Co-authored-by: Andrew Thrasher --- content/3.pecan/1.overview/2.methods-and-data.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/3.pecan/1.overview/2.methods-and-data.md b/content/3.pecan/1.overview/2.methods-and-data.md index 57291b0..f3624c4 100644 --- a/content/3.pecan/1.overview/2.methods-and-data.md +++ b/content/3.pecan/1.overview/2.methods-and-data.md @@ -153,7 +153,7 @@ Example data from 790 tumor samples can be found in [Figure 5 of McLeod et al](h ### t-SNE Methods -St. Jude Cloud's [RNA-Seq V2 WDL workflow](https://github.com/stjudecloud/workflows/blob/main/workflows/rnaseq/rnaseq-standard.wdl) was used to generate HTSeq counts data for each sample and visualized on PeCan's expression t-SNE plot. +St. Jude Cloud's [RNA-Seq V2 WDL workflow](https://github.com/stjudecloud/workflows/blob/rnaseq-standard/v2.0.0/workflows/rnaseq/rnaseq-standard.wdl) was used to generate HTSeq counts data for each sample and visualized on PeCan's expression t-SNE plot. ### t-SNE Data From ac2e11f33dfaf40d52986092402182329e35c572 Mon Sep 17 00:00:00 2001 From: Jobin Sunny Date: Tue, 17 Jun 2025 14:51:58 -0500 Subject: [PATCH 5/5] docs: use markdown footnotes for MoR paper reference --- content/3.pecan/1.overview/2.methods-and-data.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/content/3.pecan/1.overview/2.methods-and-data.md b/content/3.pecan/1.overview/2.methods-and-data.md index f3624c4..4b53154 100644 --- a/content/3.pecan/1.overview/2.methods-and-data.md +++ b/content/3.pecan/1.overview/2.methods-and-data.md @@ -166,7 +166,7 @@ Example data from 1,574 RNA-Seq samples can be accessed [here](https://viz.stjud ### Gene Expression Methods -DESeq2's Median of Ratios1 was applied to ~5500 tumor samples for gene expression analysis. +DESeq2's Median of Ratios[^1] was applied to ~5500 tumor samples for gene expression analysis. We advocate for this methodology over TPM/CPM due to its stability and resistance to outliers. #### Key Benefits @@ -176,7 +176,7 @@ We advocate for this methodology over TPM/CPM due to its stability and resistanc 3. More reliable differential expression results due to geometric mean normalization. 4. Consistent and comparable expression values across samples. -1 Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. [https://doi.org/10.1186/s13059-014-0550-8](https://doi.org/10.1186/s13059-014-0550-8) +[^1]: Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. [https://doi.org/10.1186/s13059-014-0550-8](https://doi.org/10.1186/s13059-014-0550-8) ---