<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8" />
<meta name="generator" content="pandoc" />
<meta http-equiv="X-UA-Compatible" content="IE=EDGE" />
<title>Niche Clustering</title>
<script src="site_libs/header-attrs-2.29/header-attrs.js"></script>
<script src="site_libs/jquery-3.6.0/jquery-3.6.0.min.js"></script>
<meta name="viewport" content="width=device-width, initial-scale=1" />
<link href="site_libs/bootstrap-3.3.5/css/flatly.min.css" rel="stylesheet" />
<script src="site_libs/bootstrap-3.3.5/js/bootstrap.min.js"></script>
<script src="site_libs/bootstrap-3.3.5/shim/html5shiv.min.js"></script>
<script src="site_libs/bootstrap-3.3.5/shim/respond.min.js"></script>
<style>h1 {font-size: 34px;}
h1.title {font-size: 38px;}
h2 {font-size: 30px;}
h3 {font-size: 24px;}
h4 {font-size: 18px;}
h5 {font-size: 16px;}
h6 {font-size: 12px;}
code {color: inherit; background-color: rgba(0, 0, 0, 0.04);}
pre:not([class]) { background-color: white }</style>
<script src="site_libs/jqueryui-1.13.2/jquery-ui.min.js"></script>
<link href="site_libs/tocify-1.9.1/jquery.tocify.css" rel="stylesheet" />
<script src="site_libs/tocify-1.9.1/jquery.tocify.js"></script>
<script src="site_libs/navigation-1.1/tabsets.js"></script>
<link href="site_libs/highlightjs-9.12.0/textmate.css" rel="stylesheet" />
<script src="site_libs/highlightjs-9.12.0/highlight.js"></script>
<link href="site_libs/font-awesome-6.5.2/css/all.min.css" rel="stylesheet" />
<link href="site_libs/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet" />
<style type="text/css">
code{white-space: pre-wrap;}
span.smallcaps{font-variant: small-caps;}
span.underline{text-decoration: underline;}
div.column{display: inline-block; vertical-align: top; width: 50%;}
div.hanging-indent{margin-left: 1.5em; text-indent: -1.5em;}
ul.task-list{list-style: none;}
</style>
<style type="text/css">code{white-space: pre;}</style>
<script type="text/javascript">
if (window.hljs) {
hljs.configure({languages: []});
hljs.initHighlightingOnLoad();
if (document.readyState && document.readyState === "complete") {
window.setTimeout(function() { hljs.initHighlighting(); }, 0);
}
}
</script>
<style type = "text/css">
.main-container {
max-width: 940px;
margin-left: auto;
margin-right: auto;
}
img {
max-width:100%;
}
.tabbed-pane {
padding-top: 12px;
}
.html-widget {
margin-bottom: 20px;
}
button.code-folding-btn:focus {
outline: none;
}
summary {
display: list-item;
}
details > summary > p:only-child {
display: inline;
}
pre code {
padding: 0;
}
</style>
<style type="text/css">
.dropdown-submenu {
position: relative;
}
.dropdown-submenu>.dropdown-menu {
top: 0;
left: 100%;
margin-top: -6px;
margin-left: -1px;
border-radius: 0 6px 6px 6px;
}
.dropdown-submenu:hover>.dropdown-menu {
display: block;
}
.dropdown-submenu>a:after {
display: block;
content: " ";
float: right;
width: 0;
height: 0;
border-color: transparent;
border-style: solid;
border-width: 5px 0 5px 5px;
border-left-color: #cccccc;
margin-top: 5px;
margin-right: -10px;
}
.dropdown-submenu:hover>a:after {
border-left-color: #adb5bd;
}
.dropdown-submenu.pull-left {
float: none;
}
.dropdown-submenu.pull-left>.dropdown-menu {
left: -100%;
margin-left: 10px;
border-radius: 6px 0 6px 6px;
}
</style>
<script type="text/javascript">
// manage active state of menu based on current page
$(document).ready(function () {
// active menu anchor
href = window.location.pathname
href = href.substr(href.lastIndexOf('/') + 1)
if (href === "")
href = "index.html";
var menuAnchor = $('a[href="' + href + '"]');
// mark the anchor link active (and if it's in a dropdown, also mark that active)
var dropdown = menuAnchor.closest('li.dropdown');
if (window.bootstrap) { // Bootstrap 4+
menuAnchor.addClass('active');
dropdown.find('> .dropdown-toggle').addClass('active');
} else { // Bootstrap 3
menuAnchor.parent().addClass('active');
dropdown.addClass('active');
}
// Navbar adjustments
var navHeight = $(".navbar").first().height() + 15;
var style = document.createElement('style');
var pt = "padding-top: " + navHeight + "px; ";
var mt = "margin-top: -" + navHeight + "px; ";
var css = "";
// offset scroll position for anchor links (for fixed navbar)
for (var i = 1; i <= 6; i++) {
css += ".section h" + i + "{ " + pt + mt + "}\n";
}
style.innerHTML = "body {" + pt + "padding-bottom: 40px; }\n" + css;
document.head.appendChild(style);
});
</script>
<!-- tabsets -->
<style type="text/css">
.tabset-dropdown > .nav-tabs {
display: inline-table;
max-height: 500px;
min-height: 44px;
overflow-y: auto;
border: 1px solid #ddd;
border-radius: 4px;
}
.tabset-dropdown > .nav-tabs > li.active:before, .tabset-dropdown > .nav-tabs.nav-tabs-open:before {
content: "\e259";
font-family: 'Glyphicons Halflings';
display: inline-block;
padding: 10px;
border-right: 1px solid #ddd;
}
.tabset-dropdown > .nav-tabs.nav-tabs-open > li.active:before {
content: "\e258";
font-family: 'Glyphicons Halflings';
border: none;
}
.tabset-dropdown > .nav-tabs > li.active {
display: block;
}
.tabset-dropdown > .nav-tabs > li > a,
.tabset-dropdown > .nav-tabs > li > a:focus,
.tabset-dropdown > .nav-tabs > li > a:hover {
border: none;
display: inline-block;
border-radius: 4px;
background-color: transparent;
}
.tabset-dropdown > .nav-tabs.nav-tabs-open > li {
display: block;
float: none;
}
.tabset-dropdown > .nav-tabs > li {
display: none;
}
</style>
<!-- code folding -->
<style type="text/css">
#TOC {
margin: 25px 0px 20px 0px;
}
@media (max-width: 768px) {
#TOC {
position: relative;
width: 100%;
}
}
@media print {
.toc-content {
/* see https://github.com/w3c/csswg-drafts/issues/4434 */
float: right;
}
}
.toc-content {
padding-left: 30px;
padding-right: 40px;
}
div.main-container {
max-width: 1200px;
}
div.tocify {
width: 20%;
max-width: 260px;
max-height: 85%;
}
@media (min-width: 768px) and (max-width: 991px) {
div.tocify {
width: 25%;
}
}
@media (max-width: 767px) {
div.tocify {
width: 100%;
max-width: none;
}
}
.tocify ul, .tocify li {
line-height: 20px;
}
.tocify-subheader .tocify-item {
font-size: 0.90em;
}
.tocify .list-group-item {
border-radius: 0px;
}
.tocify-subheader {
display: inline;
}
.tocify-subheader .tocify-item {
font-size: 0.95em;
}
</style>
</head>
<body>
<div class="container-fluid main-container">
<!-- setup 3col/9col grid for toc_float and main content -->
<div class="row">
<div class="col-xs-12 col-sm-4 col-md-3">
<div id="TOC" class="tocify">
</div>
</div>
<div class="toc-content col-xs-12 col-sm-8 col-md-9">
<div class="navbar navbar-default navbar-fixed-top" role="navigation">
<div class="container">
<div class="navbar-header">
<button type="button" class="navbar-toggle collapsed" data-toggle="collapse" data-bs-toggle="collapse" data-target="#navbar" data-bs-target="#navbar">
<span class="icon-bar"></span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
</button>
<a class="navbar-brand" href="index.html">VoltRon</a>
</div>
<div id="navbar" class="navbar-collapse collapse">
<ul class="nav navbar-nav">
<li>
<a href="tutorials.html">Explore</a>
</li>
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">
Vignette
<span class="caret"></span>
</a>
<ul class="dropdown-menu" role="menu">
<li class="dropdown-submenu">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">Spatial Data Integration</a>
<ul class="dropdown-menu" role="menu">
<li>
<a href="registration.html">Spatial Data Alignment</a>
</li>
<li>
<a href="multiomic.html">Multi-omic Integration</a>
</li>
<li>
<a href="nicheclustering.html">Niche Clustering</a>
</li>
</ul>
</li>
<li class="dropdown-submenu">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">Downstream Analysis</a>
<ul class="dropdown-menu" role="menu">
<li>
<a href="roianalysis.html">ROI Analysis</a>
</li>
<li>
<a href="spotanalysis.html">Cell/Spot Analysis</a>
</li>
<li>
<a href="moleculeanalysis.html">Molecule Analysis</a>
</li>
<li>
<a href="pixelanalysis.html">Pixels (Image Only) Analysis</a>
</li>
</ul>
</li>
<li class="dropdown-submenu">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">Utilities</a>
<ul class="dropdown-menu" role="menu">
<li>
<a href="interactive.html">Interactive Utilities</a>
</li>
<li>
<a href="importingdata.html">Importing Spatial Data</a>
</li>
<li>
<a href="voltronobjects.html">Working with VoltRon Objects</a>
</li>
<li>
<a href="conversion.html">Converting VoltRon Objects</a>
</li>
<li>
<a href="ondisk.html">OnDisk-based Analysis Utilities</a>
</li>
</ul>
</li>
</ul>
</li>
</ul>
<ul class="nav navbar-nav navbar-right">
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">
<span class="fa fa-envelope-o"></span>
Contact
<span class="caret"></span>
</a>
<ul class="dropdown-menu" role="menu">
<li>
<a href="https://bioinformatics.mdc-berlin.de">Altuna Lab/BIMSB Bioinfo</a>
</li>
<li>
<a href="https://www.mdc-berlin.de/landthaler">Landthaler Lab/BIMSB</a>
</li>
</ul>
</li>
<li class="dropdown">
<a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" data-bs-toggle="dropdown" aria-expanded="false">
<span class="fa fa-github"></span>
GitHub
<span class="caret"></span>
</a>
<ul class="dropdown-menu" role="menu">
<li>
<a href="https://github.com/BIMSBbioinfo/VoltRon">VoltRon</a>
</li>
<li>
<a href="https://github.com/BIMSBbioinfo">BIMSB Bioinfo</a>
</li>
</ul>
</li>
</ul>
</div><!--/.nav-collapse -->
</div><!--/.container -->
</div><!--/.navbar -->
<div id="header">
<h1 class="title toc-ignore">Niche Clustering</h1>
</div>
<style>
.title{
display: none;
}
body {
text-align: justify
}
.center {
display: block;
margin-left: auto;
margin-right: auto;
}
</style>
<style type="text/css">
.watch-out {
color: black;
}
</style>
<p><br></p>
<div id="spot-based-niche-clustering" class="section level1">
<h1>Spot-based Niche Clustering</h1>
<p>Spot-based spatial transcriptomic assays capture spatially-resolved
gene expression profiles that are somewhat closer to single cell
resolution. However, each spot still includes a few number of cells that
are likely originated from few number of cell types, hence
transcriptomic profile of each spot would likely include markers from
multiple cell types. Here, <strong>RNA deconvolution</strong> can be
incorporated to estimate the percentage/abundance of cell types for each
spot. We use a scRNAseq dataset as a reference to computationally
estimate the relative abundance of cell types across across the
spots.</p>
<p>VoltRon includes wrapper commands for using popular spot-level RNA
deconvolution methods such as <a
href="https://www.nature.com/articles/s41587-021-00830-w">RCTD</a> and
return estimated abundances as additional feature sets within each
layer. These estimated percentages of cell types for each spot could be
incorporated to detect <strong>niches</strong> (i.e. small local
microenvironments of cells) within the tissue. We can process cell type
abundance assays and used them for clustering to detect these
niches.</p>
<p><br></p>
<div id="import-visium-data" class="section level2">
<h2>Import Visium Data</h2>
<p>For this tutorial we will analyze spot-based transcriptomic assays
from Mouse Brain generated by the <a
href="https://www.10xgenomics.com/products/spatial-gene-expression">Visium</a>
instrument.</p>
<p>You can find and download readouts of all four Visium sections <a
href="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Cellanalysis/Visium/MouseBrainSerialSections.zip">here</a>.
The <strong>Mouse Brain Serial Section 1/2</strong> datasets can be also
downloaded from <a
href="https://www.10xgenomics.com/resources/datasets?menu%5Bproducts.name%5D=Spatial%20Gene%20Expression&query=&page=1&configure%5BhitsPerPage%5D=50&configure%5BmaxValuesPerFacet%5D=1000">here</a>
(specifically, please filter for <strong>Species=Mouse</strong>,
<strong>AnatomicalEntity=brain</strong>, <strong>Chemistry=v1</strong>
and <strong>PipelineVersion=v1.1.0</strong>).</p>
<p>We will now import each of four samples separately and merge them
into one VoltRon object. There are four brain tissue sections in total
given two serial anterior and serial posterior sections, hence we have
<strong>two tissue blocks each having two layers</strong>.</p>
<pre class="r watch-out"><code>library(VoltRon)
Ant_Sec1 <- importVisium("Sagittal_Anterior/Section1/", sample_name = "Anterior1")
Ant_Sec2 <- importVisium("Sagittal_Anterior/Section2/", sample_name = "Anterior2")
Pos_Sec1 <- importVisium("Sagittal_Posterior/Section1/", sample_name = "Posterior1")
Pos_Sec2 <- importVisium("Sagittal_Posterior/Section2/", sample_name = "Posterior2")
# merge datasets
MBrain_Sec_list <- list(Ant_Sec1, Ant_Sec2, Pos_Sec1, Pos_Sec2)
MBrain_Sec <- merge(MBrain_Sec_list[[1]], MBrain_Sec_list[-1],
samples = c("Anterior", "Anterior", "Posterior", "Posterior"))
MBrain_Sec</code></pre>
<pre><code>VoltRon Object
Anterior:
Layers: Section1 Section2
Posterior:
Layers: Section1 Section2
Assays: Visium(Main)
Features: RNA(Main) </code></pre>
<p>VoltRon maps metadata features on the spatial images, multiple
features can be provided for all assays/layers associated with the main
assay (Visium).</p>
<pre class="r watch-out"><code>vrSpatialFeaturePlot(MBrain_Sec, features = "Count", crop = TRUE, alpha = 1, ncol = 2)</code></pre>
<p><img width="80%" height="80%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_first_plot.png" class="center"></p>
<p><br></p>
</div>
<div id="import-scrna-data" class="section level2">
<h2>Import scRNA data</h2>
<p>We will now import the scRNA data for reference which can be
downloaded from <a
href="https://www.dropbox.com/s/cuowvm4vrf65pvq/allen_cortex.rds?dl=1">here</a>.
Specifically, we will use a scRNA data of Mouse cortical adult brain
with 14,000 cells, generated with the SMART-Seq2 protocol, from the
Allen Institute. This scRNA data is also used by the Spatial Data
Analysis tutorial in <a
href="https://satijalab.org/seurat/articles/spatial_vignette.html#integration-with-single-cell-data">Seurat</a>
website.</p>
<pre class="r watch-out"><code># install packages if necessary
if(!requireNamespace("Seurat"))
install.packages("Seurat")
if(!requireNamespace("dplyr"))
install.packages("dplyr")
# import scRNA data
library(Seurat)
allen_reference <- readRDS("allen_cortex.rds")
# process and reduce dimensionality
library(dplyr)
allen_reference <- SCTransform(allen_reference, ncells = 3000, verbose = FALSE) %>%
RunPCA(verbose = FALSE) %>%
RunUMAP(dims = 1:30)</code></pre>
<p>Before deconvoluting Visium spots, we correct cell types labels and
drop some cell types with extremely few number of cells (e.g. “CR”).</p>
<pre class="r watch-out"><code># update labels and subset
allen_reference$subclass <- gsub("L2/3 IT", "L23 IT", allen_reference$subclass)
allen_reference <- allen_reference[,colnames(allen_reference)[!allen_reference@meta.data$subclass %in% "CR"]]
# visualize
Idents(allen_reference) <- "subclass"
gsubclass <- DimPlot(allen_reference, reduction = "umap", label = T) + NoLegend()
Idents(allen_reference) <- "class"
gclass <- DimPlot(allen_reference, reduction = "umap", label = T) + NoLegend()
gsubclass | gclass</code></pre>
<p><img width="95%" height="95%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_singlecell.png" class="center"></p>
<p><br></p>
</div>
<div id="spot-deconvolution-with-rctd" class="section level2">
<h2>Spot Deconvolution with RCTD</h2>
<p>In order to integrate the scRNA data and the spatial data sets within
the VoltRon object and estimate relative cell type abundances for each
Visium spot, we will use <strong>RCTD</strong> algorithm which is
accessible with the <a
href="https://github.com/dmcable/spacexr">spacexr</a> package.</p>
<pre class="r watch-out"><code>if(!requireNamespace("spacexr"))
devtools::install_github("dmcable/spacexr", build_vignettes = FALSE)</code></pre>
<p>After running <strong>getDeconvolution</strong>, an additional
feature set within the same Visium assay with name
<strong>Decon</strong> will be created.</p>
<pre class="r watch-out"><code>library(spacexr)
MBrain_Sec <- getDeconvolution(MBrain_Sec, sc.object = allen_reference, sc.cluster = "subclass", max_cores = 6)
MBrain_Sec</code></pre>
<pre><code>VoltRon Object
Anterior:
Layers: Section1 Section2
Posterior:
Layers: Section1 Section2
Assays: Visium(Main)
Features: RNA(Main) Decon </code></pre>
<p>We can now switch to the <strong>Decon</strong> feature type where
features are cell types from the scRNA reference and the data values are
cell types percentages in each spot.</p>
<pre class="r watch-out"><code>vrMainFeatureType(MBrain_Sec) <- "Decon"
vrFeatures(MBrain_Sec)</code></pre>
<pre><code> [1] "Astro" "Endo" "L23 IT" "L4" "L5 IT" "L5 PT"
[7] "L6 CT" "L6 IT" "L6b" "Lamp5" "Macrophage" "Meis2"
[13] "NP" "Oligo" "Peri" "Pvalb" "Serpinf1" "SMC"
[19] "Sncg" "Sst" "Vip" "VLMC" </code></pre>
<p>These features (i.e. cell type abundances) can be visualized like any
other feature type.</p>
<pre class="r watch-out"><code>vrSpatialFeaturePlot(MBrain_Sec, features = c("L4", "L5 PT", "Oligo", "Vip"),
crop = TRUE, ncol = 2, alpha = 1, keep.scale = "all")</code></pre>
<p><img width="90%" height="90%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_spatialfeature_plot.png" class="center"></p>
<p><br></p>
</div>
<div id="clustering" class="section level2">
<h2>Clustering</h2>
<p>Relative cell type abundances that are learned by RCTD and stored
within VoltRon can now be used to cluster spots. These groups or
clusters of spots can often be referred to as <strong>niches</strong>.
Here, as a definition, a niche is a region or a collection of regions
within tissue that have a distinct cell type composition as opposed to
the remaining parts of the tissue.</p>
<p>The cell type abundances (which adds up to one for each spot) can be
normalized and processed like transcriptomic and proteomic profiles
prior to clustering (i.e. niche clustering). We treat cell type
abundances as <a
href="https://en.wikipedia.org/wiki/Compositional_data">compositional
data</a>, hence we incorporate <strong>centred log ratio (CLR)</strong>
transformation for normalizing them.</p>
<pre class="r watch-out"><code>vrMainFeatureType(MBrain_Sec) <- "Decon"
MBrain_Sec <- normalizeData(MBrain_Sec, method = "CLR")</code></pre>
<p>The CLR normalized assay have only 25 features, each representing a
cell type from the single cell reference data. Hence, we can
<strong>directly calculate UMAP reductions from this feature
abundances</strong> since we dont have much number of features which
necessitates dimensionality reduction such as PCA.</p>
<p>However, we may still need to reduce the dimensionality of this space
with 25 features using UMAP for visualizing purposes. VoltRon is also
capable of calculating the UMAP reduction from normalized data slots.
Hence, we build a UMAP reduction from CLR data directly. However, UMAP
will always be calculated from a PCA reduction by default (if a PCA
embedding is found in the object).</p>
<pre class="r watch-out"><code>MBrain_Sec <- getUMAP(MBrain_Sec, data.type = "norm")
vrEmbeddingPlot(MBrain_Sec, embedding = "umap", group.by = "Sample")</code></pre>
<p><img width="60%" height="60%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_embedding_sample.png" class="center"></p>
<p><br></p>
<p>Using normalized cell type abundances, we can now generate k-nearest
neighbor graphs and cluster the graph using leiden method.</p>
<pre class="r watch-out"><code>MBrain_Sec <- getProfileNeighbors(MBrain_Sec, data.type = "norm", method = "SNN")
vrGraphNames(MBrain_Sec)</code></pre>
<pre><code>[1] "SNN"</code></pre>
<pre class="r watch-out"><code>MBrain_Sec <- getClusters(MBrain_Sec, resolution = 0.6, graph = "SNN")</code></pre>
<p><br></p>
</div>
<div id="visualization" class="section level2">
<h2>Visualization</h2>
<p>VoltRon incorporates distinct plotting functions for,
e.g. embeddings, coordinates, heatmap and even barplots. We can now map
the clusters we have generated on UMAP embeddings.</p>
<pre class="r watch-out"><code># visualize
g1 <- vrEmbeddingPlot(MBrain_Sec, embedding = "umap", group.by = "Sample")
g2 <- vrEmbeddingPlot(MBrain_Sec, embedding = "umap", group.by = "niche_clusters", label = TRUE)
g1 | g2</code></pre>
<p><img width="100%" height="100%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_embedding_clusters.png" class="center"></p>
<p><br></p>
<p>Mapping clusters on the spatial images and spots would show the niche
structure across all four tissue sections.</p>
<pre class="r watch-out"><code>vrSpatialPlot(MBrain_Sec, group.by = "niche_clusters", crop = TRUE, alpha = 1)</code></pre>
<p><img width="80%" height="80%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_spatial_clusters.png" class="center"></p>
<p><br></p>
<p>We use <strong>vrHeatmapPlot</strong> to investigate relative cell
type abundances across these niche clusters. You will need to have
<strong>ComplexHeatmap</strong> package in your namespace.</p>
<pre class="r watch-out"><code># install packages if necessary
if(!requireNamespace("ComplexHeatmap"))
BiocManager::install("ComplexHeatmap")
# heatmap of niches
library(ComplexHeatmap)
vrHeatmapPlot(MBrain_Sec, features = vrFeatures(MBrain_Sec), group.by = "niche_clusters",
show_row_names = T, show_heatmap_legend = T)</code></pre>
<p><img width="90%" height="90%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_heatmap_clusters.png" class="center">
<br></p>
</div>
</div>
<div id="cell-based-niche-clustering" class="section level1">
<h1>Cell-based Niche Clustering</h1>
<p>Similar to spot-based spatial omics assays, we can build and cluster
niche associated to each cell for spatial transcriptomics datasets in
single cell resolution. For this, we require building niche assays for
the collections of cells where a niche of cell is defined as a region of
sets of regions with distinct cell type population that each of these
cells belong to.</p>
<p>Here, we dont require any scRNA reference dataset but we may first
need to cluster and annotate cells in the RNA/transcriptome level
profiles, and determine cell types. Then, we first detect the mixture of
cell types within a spatial neighborhood around all cells and use that
as a profile to perform clustering where these clusters will be
associated with niches.</p>
<div id="import-xenium-data" class="section level2">
<h2>Import Xenium Data</h2>
<p>For this, the data has to be already clustered (and annotated if
possible). We will use the cluster labels generated at the end of the
Xenium analysis workflow from <a href="spotanalysis.html">Cell/Spot
Analysis</a>. You can download the VoltRon object with clustered and
annotated Xenium cells along with the Visium assay from <a
href="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/SpatialDataAlignment/Xenium_vs_Visium/VRBlock_data_clustered.rds">here</a>.</p>
<pre class="r watch-out"><code>Xen_data <- readRDS("VRBlock_data_clustered.rds")</code></pre>
<p>We will use all these 18 cell types used for annotating Xenium cells
for detecting niches with distinct cellular type mixtures.</p>
<pre class="r watch-out"><code>vrMainSpatial(Xen_data, assay = "Assay1") <- "main"
vrMainSpatial(Xen_data, assay = "Assay3") <- "main"
vrSpatialPlot(Xen_data, group.by = "CellType", pt.size = 0.13, background.color = "black",
legend.loc = "top", n.tile = 500)</code></pre>
<p><img width="100%" height="100%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_xenium_clusters.png" class="center"></p>
<p><br></p>
</div>
<div id="creating-niche-assay" class="section level2">
<h2>Creating Niche Assay</h2>
<p>For calculating niche profiles for each cell, we have to first build
spatial neighborhoods around cells and capture the local cell type
mixtures. Using <strong>getSpatialNeighbors</strong>, we build a spatial
neighborhood graph to connect all cells to other cells within at most 15
distance apart.</p>
<pre class="r watch-out"><code>Xen_data <- getSpatialNeighbors(Xen_data, radius = 15, method = "radius")
vrGraphNames(Xen_data)</code></pre>
<pre><code>[1] "radius"</code></pre>
<p>Now, we can build a niche assay for cells using the
<strong>getNicheAssay</strong> function which will create an additional
feature set for cells called <strong>Niche</strong>. Here, each cell
type is a feature and the profile of a cell represents the relative
abundance of cell types around each cell.</p>
<pre class="r watch-out"><code>Xen_data <- getNicheAssay(Xen_data, label = "CellType", graph.type = "radius")
Xen_data</code></pre>
<pre><code>VoltRon Object
10XBlock:
Layers: Section1 Section2 Section3
Assays: Xenium(Main) Visium
Features: RNA(Main) Niche</code></pre>
<p><br></p>
</div>
<div id="clustering-1" class="section level2">
<h2>Clustering</h2>
<p>The Niche assay can be normalized similar to the spot-level niche
analysis using <strong>centred log ratio (CLR)</strong>
transformation.</p>
<pre class="r watch-out"><code>vrMainFeatureType(Xen_data) <- "Niche"
Xen_data <- normalizeData(Xen_data, method = "CLR")</code></pre>
<p>Default clustering functions could be used to analyze the normalized
niche profiles of cells to detect niches associated with each cell.
However, we use K-means algorithm to perform the niche clustering. For
this exercise, we pick an estimate of 7 clusters which will be the
number of niche clusters we get.</p>
<pre class="r watch-out"><code>Xen_data <- getClusters(Xen_data, nclus = 7, method = "kmeans", label = "niche_clusters")</code></pre>
<p>After the niche clustering, the metadata is updated and observed
later like below.</p>
<pre class="r watch-out"><code>head(Metadata(Xen_data))</code></pre>
<div>
<pre><code style="font-size: 10px;"> id Count assay_id Assay Layer Sample CellType niche_clusters
1_Assay1 1_Assay1 28 Assay1 Xenium Section1 10XBlock DCIS_1 2
2_Assay1 2_Assay1 94 Assay1 Xenium Section1 10XBlock DCIS_2 2
3_Assay1 3_Assay1 9 Assay1 Xenium Section1 10XBlock DCIS_1 2
4_Assay1 4_Assay1 11 Assay1 Xenium Section1 10XBlock DCIS_1 2
5_Assay1 5_Assay1 48 Assay1 Xenium Section1 10XBlock DCIS_2 2
6_Assay1 6_Assay1 7 Assay1 Xenium Section1 10XBlock DCIS_1 2</code></pre>
</div>
<p><br></p>
</div>
<div id="visualization-1" class="section level2">
<h2>Visualization</h2>
<p>After niche clustering, each cell in the Xenium assay will be
assigned a niche which is initially a number which indicates the ID of
each particular niche. It is up to the user to annotate, filter and
visualize these niches moving forward.</p>
<pre class="r watch-out"><code>vrSpatialPlot(Xen_data, group.by = "niche_clusters", alpha = 1, legend.loc = "top")</code></pre>
<p>We use <strong>vrHeatmapPlot</strong> to investigate the abundance of
each cell type across the niche clusters. You will need to have
<strong>ComplexHeatmap</strong> package in your namespace. We see that
niche cluster 1 include all invasive tumor subtypes (IT 1-3). We see
this for two subtypes of in situ ductal carcinoma (DCIS 1,2) subtypes as
well other than a third DCIS subcluster being within proximity to
myoepithelial cells. Niche cluster 6 also shows regions within the
breast cancer tissue where T cells and B cells are found together
abundantly.</p>
<p><img width="100%" height="100%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_xenium_nicheclusters.png" class="center">
<br></p>
<pre class="r watch-out"><code># install packages if necessary
if(!requireNamespace("ComplexHeatmap"))
BiocManager::install("ComplexHeatmap")
# heatmap of niches
library(ComplexHeatmap)
vrHeatmapPlot(Xen_data, features = vrFeatures(Xen_data), group.by = "niche_clusters")</code></pre>
<p><img width="100%" height="100%" src="https://bimsbstatic.mdc-berlin.de/landthaler/VoltRon/Package/images/decon_xenium_heatmapclusters.png" class="center">
<br></p>
</div>
</div>
</div>
</div>
</div>
<script>
// add bootstrap table styles to pandoc tables
function bootstrapStylePandocTables() {
$('tr.odd').parent('tbody').parent('table').addClass('table table-condensed');
}
$(document).ready(function () {
bootstrapStylePandocTables();
});
</script>
<!-- tabsets -->
<script>
$(document).ready(function () {
window.buildTabsets("TOC");
});
$(document).ready(function () {
$('.tabset-dropdown > .nav-tabs > li').click(function () {
$(this).parent().toggleClass('nav-tabs-open');
});
});
</script>
<!-- code folding -->
<script>
$(document).ready(function () {
// temporarily add toc-ignore selector to headers for the consistency with Pandoc
$('.unlisted.unnumbered').addClass('toc-ignore')
// move toc-ignore selectors from section div to header
$('div.section.toc-ignore')
.removeClass('toc-ignore')
.children('h1,h2,h3,h4,h5').addClass('toc-ignore');
// establish options
var options = {
selectors: "h1,h2,h3",
theme: "bootstrap3",
context: '.toc-content',
hashGenerator: function (text) {
return text.replace(/[.\\/?&!#<>]/g, '').replace(/\s/g, '_');
},
ignoreSelector: ".toc-ignore",
scrollTo: 0
};
options.showAndHide = false;
options.smoothScroll = false;
// tocify
var toc = $("#TOC").tocify(options).data("toc-tocify");
});
</script>
<!-- dynamically load mathjax for compatibility with self-contained -->
<script>
(function () {
var script = document.createElement("script");
script.type = "text/javascript";
script.src = "https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML";
document.getElementsByTagName("head")[0].appendChild(script);
})();
</script>
</body>
</html>
Datasets
Models
