CosMx Analysis Scratch Space – Blog

Creating reference profiles for InSituType

cell typing

A short guide to creating reference profiles, and to merging profiles from different sources.

Patrick Danaher

How does background impact CosMx data, and when does it matter?

quality control

overview

CosMx data includes off-target “background” counts from off-target probe binding, appearing randomly at a low frequency across the dataset. In this article, we’ll discuss…

Patrick Danaher

InSituCor: an open-source package and AtoMx custom module for exploratory data analysis

AtoMx modules

exploratory analysis

We developed InSituCor as a tool for exploratory analysis of CosMx data; it has since become a mainstay of our in-house analyses. InSituCor reports on modules of “spatially…

Patrick Danaher

Comparing AtoMx™ SIP flat files with the legacy format

flat files

This short post shows differences between the AtoMx™ SIP-exported and legacy CosMx™ SMI "flat files"

Using Squidpy with AtoMx™ SIP exports

Squidpy

python

visualization

AnnData

In this blog post, I’ll show how to prepare and analyze AtoMx™ SIP-exported CosMx™ SMI data with python's squidpy package.

Applications for visualization and cell typing using ‘smoothed’ marker genes

visualization

cell typing

Applications for visualization and cell typing using ‘smoothed’ marker genes based on expression nearest neighbors

`napari-cosmx` essentials

napari

how-tos

visualization

In this post, I walk through some of the basic ways I use napari-cosmx to view and analyze SMI data. I will make use of this GUI/scripting duality and share a combination of GUI and programmatic tips and tricks.

GeoMx® mask generation for marker-based single-cell application

GeoMx

how-tos

image processing

Python

This post introduces a pipeline for automatic generating GeoMx® Digital Spatial Profiler (DSP)-ready binary masks in batch for marker-based single-cell application. Given the query marker protein of interest, the pipeline would take morphology images and generate binary masks for negative-stained cells and cells connecting to positive-stained regions, respectively. The pipeline runs as a command line and this post would serve as a guide to how it works and how to use it.

Making composite images from layered morphology 2D images

Squidpy

Giotto

pre-processing

python

In this post, we describe a developmental python script that creates composite images from layered morphology 2D images.

Evelyn Metzger, Vikram Kohli

Creating an anndata object from AtoMx™ exported results for interactive viewer and python-based analysis

how-tos

visualization

Seurat

AnnData

python

This post describes how to create anndata object from AtoMx™ exported results. The resulting oject in ‘.h5ad’ format could be further analyzed using various python-based single-cell analysis tools, like scanpy and squidpy. Non-coders could also share the light-weighted data object, visualize and explore the processed data in several open-sourced interactive viewers, like Cirrocumulus and CELLxGENE viewers.

Vignette: Basics of CosMx Analysis in R

recommended

overview

quality control

normalization

cell typing

pre-processing

A complete demo analysis of a CosMx dataset can be found at workflow section in this post and the correspond scripts are stored under _code/vignette folder in this…

Patrick Danaher

FOV QC from single-cell gene expression in spatial dataset

quality control

pre-processing

This experimental approach is new as of April 2024 and is lightly tested. Please use thoughtfully while considering biological context.

Patrick Danaher

Evaluating Cell Segmentation Error based on Transcriptional Spatial Profiles using FastReseg

segmentation

algorithms

FastReseg algorithm scores individual transcripts for the goodness-of-fit within their respective cells based on the probability of each gene belonging to each cell type and the spatial dependency of transcript score. FastReseg can flag cells with putative cell segmentation errors and perform corrections rapidly.

Introduction to visualizing CosMx SMI data in Seurat

visualization

Seurat

Recommendations for spatial plots in Seurat

Claire Williams

Getting Started with the `napari-cosmx` plugin

visualization

napari

python

The AtoMx™ Spatial Informatics Portal (SIP) is an excellent solution for analyzing multi-omics spatial data generated on the CosMx™ Spatial Molecular Imager (SMI). It offers…

The spatial algorithms zoo: recommended algorithms and efficient code

algorithms

Spatial statistics is a well-developed field, with deep statistical methodology and highly efficient open-source tools. In CosMx data, where a single study can contain…

Patrick Danaher

Cell typing: what we’ve found to work

cell typing

We cell type most studies using one of the following approaches:

Patrick Danaher

On the use of marker genes

cell typing

Our basic recommendation is this: relying on a few marker genes alone will not produce successful cell typing.

Patrick Danaher

Big datasets: strategies for memory-efficient analysis

big data

CosMx data can be truly huge, containing millions of cells and thousands of genes. This prevents many typical analysis strategies, including many toolkits designed for…

Patrick Danaher

QC and normalization of RNA data

quality control

normalization

pre-processing

We’ve tried a lot of options here, and we’ve settled on very simple procedures for most cases.

Patrick Danaher

Functions for condensing FOVs and tissues to minimize whitespace

visualization

Minimizing whitespace while plotting cells in xy space is a constant challenge. A single tissue will often have discontinuous FOVs, and aligning multiple tissues in a…

Patrick Danaher

Visualize cellular neighborhood in gallery mode

visualization

napari

A complete CosMx dataset will contain cell metadata, morphology/protein images and cell label results of cell segmentation. We’ve created a toolkit for visualizing the…

Lidan Wu, Patrick Danaher

The impact of segmentation error on differential expression analyses

segmentation

Cell segmentation is always imperfect, leaving some cells’ expression profiles contaminated with transcripts properly belonging to other cells. For many analyses, this is a…

Patrick Danaher

What is high-plex spatial transcriptomics for?

overview

High-plex single cell spatial transcriptomics data is a little awe-inspiring: even a perfunctory analysis of a single run produces a terabyte of data, gorgeous images, and…

Patrick Danaher

Inferring cell polygons from transcript locations

visualization

A complete CosMx dataset will contain polygonal boundaries for each cell for use in plotting. In practice, especially with earlier datasets or with datasets passed between…

Patrick Danaher