ssHMRF is an R package built on top of smFishHmrf to allow easy identification of spatial domains from spatial transcriptomics data using hidden markov random fields.
Below we describe an example to use the package using a sample dataset provided with the package:
expression_filepath <- "fcortex.expression.txt"
coordinates_filepath <- "fcortex.coordinates.txt"
genes_filepath <- "genes"
expr <- readExpression(expression_filepath)
coords <- readCoordinates(coordinates_filepath)
fields <- coords$field
coords <- coords$coords
genes <- readGenes(genes_filepath)
spatial_network <- createSpatialNetwork(coords = coords, k = 4)
hmrfParams <- initializeHMRF(expr = expr,
spatial_network = spatial_network,
spatial_genes = genes,
k = 3)
hmrfResults <- runHMRF(expr = expr,
nei = hmrfParams$nei,
numnei = hmrfParams$numnei,
blocks = hmrfParams$blocks,
damp = hmrfParams$damp,
mu = hmrfParams$mu,
sigma = hmrfParams$sigma,
edgelist = hmrfParams$edgelist,
k = 3,
genes = genes,
beta_initial_value = 0,
beta_increment_value = 10,
beta_iterations = 5)
hmrfPlots <- plotHMRF(spatial_network = spatial_network, HMRFoutput = hmrfResults)
hmrfPlots