spatialGE.SpatialAutocorrelation
Description
Assess the level of spatial uniformity in gene expression by calculating Moran’s I and/or Geary’s C and qualitatively explore correlations with sample-level metadata (i.e., tissue type, therapy, disease status). It is used to explore the relationship between a clinical (sample-level) variable of interest and the level of gene expression spatial uniformity within a sample.
The calculation of spatial statistics uses the SThet function of spatialGE and this is followed with multi-sample comparison using spatialGE’s compare_SThet.
Module Details
- Authors: Fridley Lab of the Moffit Cancer Center
- Adapted as a GenePattern Module by: Ted Liefeld (jliefeld@cloud.ucsd.edu)
- Categories: spatial transcriptomics
- Source Repo: https://github.com/FridleyLab/spatialGE
- Module Source Repo: https://github.com/genepattern/spatialGE.SpatialAutocorrelation
- Contact: edh021@cloud.ucsd.edu
Methodology
Refer to the Fridely lab documentation for SpatialGE for original details of the method.
In order to explore the relationship between spatial heterogeneity and phenotypic or clinical data, spatialGE implemented the estimation of two autocorrelation [Moran’s I (Moran, 1950) and Geary’s C (Geary, 1954)] methods using the R package spdep (Bivand, et al., 2011). Positive autocorrelation is indicated is by a high Moran’s I (Imax=1) and low Geary’s C (Cmin=0). When positive autocorrelation is observed in the context of ST, nearby spots tend to be similar in expression of a given gene. In other words, spots with high expression of a gene are near other spots with high expression, and spots with low expression of a gene are near other spots with low expression. When Moran’s I nears zero, and Geary’s C nears 1, gene expression within an ST slice is randomly distributed and compartmentalization of gene expression is not evident. Thus, Moran’s I and Geary’s C provide a measure of spatial uniformity in gene expression, and detection of transcriptionally divergent regions within a tissue.
References
Ospina, O. E., Wilson C. M., Soupir, A. C., Berglund, A. Smalley, I., Tsai, K. Y., Fridley, B. L. 2022. spatialGE: quantification and visualization of the tumor microenvironment heterogeneity using spatial transcriptomics. Bioinformatics, 38:2645-2647. https://doi.org/10.1093/bioinformatics/btac145
Parameters
* indicates required parameter
Autocorrelation Parameters
- *stlist.rds.input (.rds)
- An STList RDS object containing normalized data generated by spatialGE.Preprocessing.
- *output filename
- Prefix for all output filenames (rds and png).
- *genes
- Comma-separated list of gene names to compute statistics. Do not include any spaces between genes and commas (e.g. “NDRG1,IGKC” not “NDRG1, IGKC”)
- samples
- Comma-separated list of sample names or indices to compute statistics. If left blank, all samples are used. Do not include any spaces between sample names and commas.
- *method
- Spatial statistic(s) to estimate: ‘moran’, ‘geary’, or both.
- *neighbors
- Number of neighbors to estimate weights.
- *overwrite
- Indicating if previous statistics (if present in the input RDS object) should be overwritten.
Sample Comparison Parameters
- samplemeta
- A string indicating the name of the variable in the clinical data frame. If NULL, uses sample names.
- *color.by
- The variable in spatial metadata used to color points in the plot. If NULL, each sample is assigned a different color
- *categorical -Logical indicating whether or not to treat color_by as a categorical variable. Default is TRUE.
- *color.palette
- A string of a color palette from khroma or RColorBrewer, or a vector with colors with enough elements to plot categories.
- *point.size
- A number specifying the size of the points in the generated plots.
Input Files
- stlist.rds.input: RDS (R Data Storage)
- Description: The input RDS object containing the bioinformatics data.
Output Files
- rds file: An updated version of the STList object provided as input, with spatial autocorrelation statistics added.
- multi-sample comparison plot(s): Plot(s) of the samplemeta versus the calculated statitsics for the named genes.
Version Comments
- 1.0.0 (2023-01-12): Initial draft of documentation.