sqsun / nnsvg

nnSVG: scalable method to identify spatially variable genes (SVGs) in spatially-resolved transcriptomics data based on nearest-neighbor Gaussian processes (NNGPs)

nnSVG

Overview

nnSVG is a method for scalable identification of spatially variable genes (SVGs) in spatially-resolved transcriptomics data.

The nnSVG method is based on nearest-neighbor Gaussian processes (Datta et al., 2016, Finley et al., 2019) and uses the BRISC algorithm (Saha and Datta, 2018) for model fitting and parameter estimation. nnSVG allows identification and ranking of SVGs with flexible length scales across a tissue slide or within spatial domains defined by covariates. The method scales linearly with the number of spatial locations and can be applied to datasets containing thousands or more spatial locations.

nnSVG is implemented as an R package within the Bioconductor framework, and is available from Bioconductor.

Our paper describing the method is available from bioRxiv.

Installation

The package can be installed from Bioconductor as follows, using R version 4.2 or above:

install.packages("BiocManager")
BiocManager::install("nnSVG")

Alternatively, the latest development version of the package can also be installed from GitHub:

remotes::install_github("lmweber/nnSVG")

If you are installing from GitHub, the following dependency packages may also need to be installed manually from Bioconductor and CRAN:

install.packages("BiocManager")
BiocManager::install("SpatialExperiment")
BiocManager::install("STexampleData")
install.packages("BRISC")

Input data format

In the examples below, we assume the input data are provided as a SpatialExperiment Bioconductor object. In this case, the outputs are stored in the rowData of the SpatialExperiment object.

However, the inputs can also be provided as a numeric matrix of normalized and transformed counts (e.g. log-transformed normalized counts) and a numeric matrix of spatial coordinates.

To provide the inputs as numeric matrices, please install the development version of the package from GitHub or the devel version of Bioconductor (which will become the new Bioconductor release version in October 2022).

Workflow

An example workflow is provided in the package vignette from Bioconductor and shown here in short form.

Load packages

library(nnSVG)
library(STexampleData)
library(scran)
library(ggplot2)

Load example dataset

# load example dataset from STexampleData package
spe <- Visium_humanDLPFC()
dim(spe)

## [1] 33538  4992

Preprocessing

# keep spots over tissue
spe <- spe[, colData(spe)$in_tissue == 1]
dim(spe)

## [1] 33538  3639

# spot-level quality control: already performed on this example dataset

# filter low-expressed and mitochondrial genes
# using function from nnSVG package with default filtering parameters
spe <- filter_genes(spe)

## Gene filtering: removing mitochondrial genes
## removed 13 mitochondrial genes
## Gene filtering: retaining genes with at least 3 counts in at least 0.5% (n = 19) of spatial locations
## removed 30216 out of 33525 genes due to low expression

# calculate log-transformed normalized counts (logcounts) using scran package
set.seed(123)
qclus <- quickCluster(spe)
spe <- computeSumFactors(spe, cluster = qclus)
spe <- logNormCounts(spe)
assayNames(spe)

## [1] "counts"    "logcounts"

Subset data for this example

# select small set of random genes and several known SVGs for faster runtime in this example workflow
set.seed(123)
ix_random <- sample(seq_len(nrow(spe)), 10)
known_genes <- c("MOBP", "PCP4", "SNAP25", "HBB", "IGKC", "NPY")
ix_known <- which(rowData(spe)$gene_name %in% known_genes)
ix <- c(ix_known, ix_random)

spe <- spe[ix, ]
dim(spe)

## [1]   16 3639

Run nnSVG

# set seed for reproducibility
# run nnSVG using a single thread for this example workflow
set.seed(123)
spe <- nnSVG(spe, n_threads = 1)

# show results
rowData(spe)

## DataFrame with 16 rows and 17 columns
## [...]

Investigate results

The results are stored in the rowData of the SpatialExperiment object.

The main results of interest are:

• LR_stat: likelihood ratio (LR) statistics used to rank SVGs
• rank: rank of top SVGs according to LR statistics
• pval: approximate p-values
• padj: approximate p-values adjusted for multiple testing
• prop_sv: effect size defined as proportion of spatial variance

# number of significant SVGs
table(rowData(spe)$padj <= 0.05)

## 
## FALSE  TRUE 
##     7     9

# show results for top n SVGs
n <- 10
rowData(spe)[order(rowData(spe)$rank)[1:n], ]

## DataFrame with 10 rows and 17 columns
##                         gene_id   gene_name    feature_type   sigma.sq
##                     <character> <character>     <character>  <numeric>
## ENSG00000168314 ENSG00000168314        MOBP Gene Expression 1.86459294
## ENSG00000132639 ENSG00000132639      SNAP25 Gene Expression 0.33829228
## ENSG00000211592 ENSG00000211592        IGKC Gene Expression 0.59161928
## ENSG00000244734 ENSG00000244734         HBB Gene Expression 0.34819123
## ENSG00000183036 ENSG00000183036        PCP4 Gene Expression 0.22354847
## ENSG00000122585 ENSG00000122585         NPY Gene Expression 0.29511061
## ENSG00000129562 ENSG00000129562        DAD1 Gene Expression 0.03687246
## ENSG00000114923 ENSG00000114923      SLC4A3 Gene Expression 0.01123674
## ENSG00000133606 ENSG00000133606       MKRN1 Gene Expression 0.00543859
## ENSG00000149923 ENSG00000149923       PPP4C Gene Expression 0.12004347
##                    tau.sq        phi    loglik   runtime      mean       var
##                 <numeric>  <numeric> <numeric> <numeric> <numeric> <numeric>
## ENSG00000168314  0.371646   0.922937  -3716.46     0.959  0.841100  1.382681
## ENSG00000132639  0.440346   3.570016  -3940.04     0.668  3.464790  0.779762
## ENSG00000211592  0.464762  20.035566  -4580.98     1.618  0.630200  1.042847
## ENSG00000244734  0.365750  27.611193  -4114.59     2.233  0.418996  0.729640
## ENSG00000183036  0.456889   8.700988  -4041.98     1.001  0.684281  0.681316
## ENSG00000122585  0.302841  68.183198  -4087.69     1.375  0.401353  0.599801
## ENSG00000129562  0.484816   8.805056  -3942.80     2.051  0.561114  0.523034
## ENSG00000114923  0.237750  16.239042  -2621.78     0.992  0.249525  0.249055
## ENSG00000133606  0.277671   0.537947  -2861.92     1.873  0.295867  0.283165
## ENSG00000149923  0.132992 198.872410  -2660.25     5.132  0.235632  0.253096
##                     spcov   prop_sv loglik_lm    LR_stat      rank        pval
##                 <numeric> <numeric> <numeric>  <numeric> <numeric>   <numeric>
## ENSG00000168314  1.623470 0.8338075  -5752.58 4072.24582         1 0.00000e+00
## ENSG00000132639  0.167868 0.4344665  -4710.39 1540.68468         2 0.00000e+00
## ENSG00000211592  1.220514 0.5600433  -5239.35 1316.74199         3 0.00000e+00
## ENSG00000244734  1.408312 0.4877029  -4589.50  949.82287         4 0.00000e+00
## ENSG00000183036  0.690958 0.3285363  -4464.82  845.69334         5 0.00000e+00
## ENSG00000122585  1.353525 0.4935363  -4232.97  290.54324         6 0.00000e+00
## ENSG00000129562  0.342215 0.0706791  -3983.78   81.97837         7 0.00000e+00
## ENSG00000114923  0.424820 0.0451298  -2633.77   23.96612         8 6.24917e-06
## ENSG00000133606  0.249257 0.0192102  -2867.31   10.77036         9 4.58402e-03
## ENSG00000149923  1.470399 0.4744140  -2663.05    5.60524        10 6.06510e-02
##                        padj
##                   <numeric>
## ENSG00000168314 0.00000e+00
## ENSG00000132639 0.00000e+00
## ENSG00000211592 0.00000e+00
## ENSG00000244734 0.00000e+00
## ENSG00000183036 0.00000e+00
## ENSG00000122585 0.00000e+00
## ENSG00000129562 0.00000e+00
## ENSG00000114923 1.24983e-05
## ENSG00000133606 8.14937e-03
## ENSG00000149923 9.70416e-02

Plot expression of top SVG

Plot expression of the top-ranked SVG.

# plot spatial expression of top-ranked SVG
ix <- which(rowData(spe)$rank == 1)
ix_name <- rowData(spe)$gene_name[ix]
ix_name

## [1] "MOBP"

df <- as.data.frame(cbind(spatialCoords(spe), expr = counts(spe)[ix, ]))

ggplot(df, aes(x = pxl_col_in_fullres, y = pxl_row_in_fullres, 
               color = expr)) + 
  geom_point(size = 0.8) + 
  coord_fixed() + 
  scale_y_reverse() + 
  scale_color_gradient(low = "gray90", high = "blue", name = "counts") + 
  ggtitle(ix_name) + 
  theme_bw() + 
  theme(panel.grid = element_blank(), 
        axis.title = element_blank(), 
        axis.text = element_blank(), 
        axis.ticks = element_blank())

Citation

Our paper describing nnSVG is available from bioRxiv:

• Weber L.M. et al. (2022), "nnSVG: scalable identification of spatially variable genes using nearest-neighbor Gaussian processes", bioRxiv