simAnn is an R package designed for semantic similarity analysis in biological ontologies, particularly Gene Ontology (GO) and KEGG pathways. It provides comprehensive methods for ontology traversal, clustering, and similarity calculations using various approaches, including information content (IC) and graph-based methods. The package facilitates ontology-based clustering and visualization, supporting both standard ontological analyses and protein complex-based clustering.
-
Multiple Ontology Support:
- Gene Ontology (GO) analysis with species-specific annotations
- KEGG pathway analysis with hierarchical relationships
- Custom ontology tree construction
- Support for different GO namespaces (BP, CC, MF)
-
Semantic Similarity Analysis:
- Term-based methods (kappa, overlap, jaccard, dice)
- Information content methods (universal, annotation-based)
- Graph-based methods (Wang's method, GOGO)
- Customizable edge weights for relationship types
- Support for new annotations and enrichment results
-
Clustering Capabilities:
- Standard term clustering based on similarity scores
- Protein complex-based clustering for GO terms
- Customizable threshold selection
- Multiple clustering methods and parameters
- Support for weighted clustering with gene sets
-
Visualization Tools:
- Network visualization of term relationships
- Hierarchical and circular tree layouts
- Customizable heatmaps with multiple color schemes
- Interactive term focusing and highlighting
- Various annotation styles and color schemes
# Install from GitHub
devtools::install_github("guokai8/simAnn")library(simAnn)
# For visualization
library(ggplot2)
library(igraph)# Build GO tree with specific namespace and species
tree <- buildGOTree(
species = "human",
namespace = "BP", # Options: "BP", "CC", "MF"
keytype = "SYMBOL", # Options: "SYMBOL", "ENTREZID"
builtin = TRUE, # Use built-in data
verbose = TRUE # Show progress messages
)
# Build KEGG tree
kegg_tree <- buildKEGGTree(
species = "human",
keytype = "SYMBOL",
builtin = TRUE
)
# Build custom ontology tree
parents <- c("a", "a", "b", "b", "b", "c", "d")
children <- c("b", "c", "c", "d", "e", "e", "f")
annotation <- list(
"a" = 1:3,
"b" = 3:4,
"c" = 5,
"d" = 7,
"e" = 4:7,
"f" = 8
)
custom_tree <- buildOntologyTree(
parentTerms = parents,
childTerms = children,
annotations = annotation
)# Different similarity methods
sim_mat <- simterm(tree,
terms = c("GO:0008150", "GO:0000011"),
method = "term")
sim_mat_resnik <- simterm(tree,
terms = 1:30,
method = "resnik",
normType = "max")weights <- c("is_a" = 0.8, "part_of" = 0.6)
sim_mat<-simterm(tree,1:50, method = "wang",weights = weights)
threshold <- pickThreshold(sim_mat,thresholds = seq(0.5,0.9,0.01),cluster_method = "components",quality_metric = "modularity")
clu <- clusterST(tree, 1:50,
method = "wang",
weights = weights,
threshold$best_threshold)# Define example gene sets
geneset <- list(
"GO:0000086" = c("TAF2", "FOXM1", "RRM1", "ATR"),
"GO:0000422" = c("LRBA", "EIF2AK1", "SPATA18"),
"GO:0001678" = c("SYBU", "KCNB1", "PLA2G6", "LIN28A"),
"GO:0001704" = c("SNAI1", "GPI", "AHDC1"),
"GO:0001707" = c("SNAI1", "GPI", "AHDC1"),
"GO:0002573" = c("IL25", "CAMK4", "EIF2AK1", "TNFRSF11B", "IL17A"),
"GO:0002761" = c("CAMK4", "TNFRSF11B", "IL17A"),
"GO:0006282" = c("TAF2", "FOXM1", "BRD8", "ATR", "PARPBP"),
"GO:0006284" = c("POLL", "APEX2"),
"GO:0006302" = c("FOXM1", "BRD8", "POLL", "ATR", "PARPBP"),
"GO:0006909" = c("IL2RB", "EIF2AK1", "PLA2G6", "TUB"),
"GO:0007498" = c("SNAI1", "GPI", "AHDC1")
)
# Define weights for different relationship types
weights <- c("is_a" = 0.8, "part_of" = 0.6)
# Calculate weighted similarity
sim_mat_weighted <- simtermW(tree,
terms = names(geneset),
new_annot = geneset,
method = "wang",
weights = weights)
# or you have the results from richR(guokai8/richR)
enrichment_df <- data.frame(
Annot = c("GO:0006915", "GO:0012501"),
GeneID = c("CASP3,BCL2", "CASP3,BAX"),
Pvalue = c(0.001, 0.002)
)
sim_mat_weighted <- simtermW(tree,
terms = enrichment_df$Annot,
new_annot = enrichment_df,
method = "wang",
weights = weights)# Find optimal threshold
threshold <- pickThreshold(
sim_mat_weighted,
thresholds = seq(0.1, 0.9, 0.01),
cluster_method = "components",
quality_metric = "modularity"
)
# Cluster with gene sets
clu <- clusterSTW(tree,
terms = names(geneset),
new_annot = geneset,
method = "wang",
weights = weights,
threshold = threshold$best_threshold)# Example enrichment data
enrichment_df <- data.frame(
Annot = c("GO:0006915", "GO:0012501"),
GeneID = c("CASP3,BCL2", "CASP3,BAX"),
Pvalue = c(0.001, 0.002)
)
# Cluster using protein complex information
complex_clusters <- clusterComplex(
enrichment_df,
species = "human",
ontology_type = "BP",
gene_delim = ",",
verbose = TRUE
)
# With enrichment results from richR
# clu <- clusterComplex(
# as.data.frame(res),
# species = "human",
# ontology_type = "BP",
# verbose = TRUE
# )# Basic tree layout
plotOntologyTree(custom_tree, layout = "tree")
# Focus on specific terms
plotOntologyTree(tree,
focus_terms = names(geneset)[1:5],
layout = "tree")# Plot cluster network
plotClusterNetwork(clu,
show_names = TRUE,
annotation_style = "combined",
layout = "fr",
vertex_size = 12)# Plot similarity heatmap
plotClusterHeatmap(clu,
show_names = TRUE,
color_scheme = "brewer",
annotation_style = "separate")# List all similarity methods
AllSimMethod()
# List all IC calculation methods
AllICMethod()For detailed function documentation and examples:
help(package = "simAnn")
?simterm # Basic similarity calculation
?simtermW # Weighted similarity calculation
?clusterST # Basic clustering
?clusterSTW # Weighted clustering
?clusterComplex # Complex-based clustering- Kai Guo ([email protected])
This package is released under the GPL_v3 License.
If you use simAnn in your research, please contact [email protected]
This project includes some modified code from the simona package.