diff --git a/paper/jats/paper.jats b/paper/jats/paper.jats index 7dcd97b..d952556 100644 --- a/paper/jats/paper.jats +++ b/paper/jats/paper.jats @@ -125,7 +125,7 @@ a Creative Commons Attribution 4.0 International License (CC BY features, including progesterone and estrogen receptor status are shown on the marginal plot at the bottom. A detailed description of the ggoncoplot sorting algorithm is available - here + here.

@@ -133,28 +133,20 @@ a Creative Commons Attribution 4.0 International License (CC BY

Example of the ggoncoplot shown in Figure 1, where the oncoplot has been dynamically cross-linked to a gene expression t-SNE plot (top left) and a methylation UMAP (top right). Here, the - lasso tool (see - [fig:lasso_select]) - was used to select a cluster of gene expression data points (i.e., - individual samples) in the t-SNE plot. Selected samples were - automatically highlighted on the UMAP and oncoplot. This reveals - that samples which cluster on the left of the t-SNE plot also - cluster in the oncoplot, chiefly containing mutations in TP53 but - wild type PIK3CA. The plots of progesterone, estrogen, HER2 status - and triple negative classification show that the samples selected in - the t-SNE are virtually all triple negative breast cancers. In - contrast to the oncoplot, the methylation UMAP shows no strong - clustering, in line with knowledge of methylation patterns in triple - negative breast cancer. + lasso tool was used to select a cluster of gene expression data + points (i.e., individual samples) in the t-SNE plot. Selected + samples were automatically highlighted on the UMAP and oncoplot. + This reveals that samples which cluster on the left of the t-SNE + plot also cluster in the oncoplot, chiefly containing mutations in + TP53 but wild type PIK3CA. The plots of progesterone, estrogen, HER2 + status and triple negative classification show that the samples + selected in the t-SNE are virtually all triple negative breast + cancers. In contrast to the oncoplot, the methylation UMAP shows no + strong clustering, in line with knowledge of methylation patterns in + triple negative breast cancer.

- -

Example image of the t-SNE plot where the lasso tool is - being used to manually delineate a data point cluster on the left - side.

- - Statement of Need diff --git a/paper/paper.md b/paper/paper.md index ba557ab..ecbc478 100644 --- a/paper/paper.md +++ b/paper/paper.md @@ -32,7 +32,7 @@ affiliations: The ggoncoplot R package generates interactive oncoplots (also called oncoprints) to visualize mutational patterns across patient cancer cohorts (\autoref{fig:oncoplot}). Oncoplots reveal patterns of gene co-mutation and include marginal plots that indicate co-occurrence of gene mutations and tumour features. It is useful to relate gene mutation patterns seen in an oncoplot to patterns seen in other plot types, including gene expression t-SNE plots or methylation UMAPs. There are, however, no existing oncoplot-generating R packages that support dynamic data linkage between different plots. To addresses this gap and enable rapid exploration of a variety of data types we constructed the ggoncoplot package for the production of oncoplots that are easily integrated with custom visualisations and that support synchronised data-selections across plots (\autoref{fig:multimodal_selection}). ggoncoplot is available on GitHub at . -![ggoncoplot output visualising mutational trends in the TCGA breast carcinoma cohort. Individual patient samples are plotted on the x-axis, hierarchically sorted so that samples with the most frequent gene mutations appear on the leftmost side. The plot indicates that PIK3CA is the most frequently mutated gene, followed by TP53. Marginal plots indicate the total number of mutations per sample (top), and the number of samples showing mutations in each gene, coloured by mutation type (right). A range of clinical features, including progesterone and estrogen receptor status are shown on the marginal plot at the bottom. A detailed description of the ggoncoplot sorting algorithm is available [here]((https://selkamand.github.io/ggoncoplot/articles/sorting_algorithm.html)) \label{fig:oncoplot}](oncoplot.pdf) +![ggoncoplot output visualising mutational trends in the TCGA breast carcinoma cohort. Individual patient samples are plotted on the x-axis, hierarchically sorted so that samples with the most frequent gene mutations appear on the leftmost side. The plot indicates that PIK3CA is the most frequently mutated gene, followed by TP53. Marginal plots indicate the total number of mutations per sample (top), and the number of samples showing mutations in each gene, coloured by mutation type (right). A range of clinical features, including progesterone and estrogen receptor status are shown on the marginal plot at the bottom. A detailed description of the ggoncoplot sorting algorithm is available [here](https://selkamand.github.io/ggoncoplot/articles/sorting_algorithm.html). \label{fig:oncoplot}](oncoplot.pdf) ![Example of the ggoncoplot shown in Figure 1, where the oncoplot has been dynamically cross-linked to a gene expression t-SNE plot (top left) and a methylation UMAP (top right). Here, the lasso tool was used to select a cluster of gene expression data points (i.e., individual samples) in the t-SNE plot. Selected samples were automatically highlighted on the UMAP and oncoplot. This reveals that samples which cluster on the left of the t-SNE plot also cluster in the oncoplot, chiefly containing mutations in TP53 but wild type PIK3CA. The plots of progesterone, estrogen, HER2 status and triple negative classification show that the samples selected in the t-SNE are virtually all triple negative breast cancers. In contrast to the oncoplot, the methylation UMAP shows no strong clustering, in line with knowledge of methylation patterns in triple negative breast cancer. \label{fig:multimodal_selection}](multimodal_selection_with_lasso.png) diff --git a/paper/paper.pdf b/paper/paper.pdf index a9d02dc..0afc399 100644 Binary files a/paper/paper.pdf and b/paper/paper.pdf differ