,
Denes Turei [aut] ,
Michael P. Menden [aut] ,
Albert Krewinkel [ctb, cph] (pagebreak Lua filter)| Title: | Weighting protein-protein interactions |
|---|---|
| Description: | Protein-protein interaction data is essential for omics data analysis and modeling. Database knowledge is general, not specific for cell type, physiological condition or any other context determining which connections are functional and contribute to the signaling. Functional annotations such as Gene Ontology and Human Phenotype Ontology might help to evaluate the relevance of interactions. This package predicts functional relevance of protein-protein interactions based on functional annotations such as Human Protein Ontology and Gene Ontology, and prioritizes genes based on network topology, functional scores and a path search algorithm. |
| Authors: | Ana Galhoz [cre, aut] ,
Denes Turei [aut] ,
Michael P. Menden [aut] ,
Albert Krewinkel [ctb, cph] (pagebreak Lua filter) |
| Maintainer: | Ana Galhoz <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 1.13.0 |
| Built: | 2024-07-02 03:17:33 UTC |
| Source: | https://github.com/bioc/wppi |
For each interacting pair of proteins in the PPI network, store the nodes of the common neighbors. This function works for any igraph graph.
common_neighbors(graph_op)common_neighbors(graph_op)
graph_op |
Igraph object based on OmniPath PPI interactions from
|
Data frame (tibble) with igraph vertex IDs of connected pairs of vertices (source and target), a list column with the IDs of their common neighbors, and a column with the number of neighbors.
graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) shared_neighbors <- common_neighbors(graph_op_1)graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) shared_neighbors <- common_neighbors(graph_op_1)
Number of total genes in an ontology database
count_genes(data_annot)count_genes(data_annot)
data_annot |
Data frame (tibble) of GO or HPO datasets from
|
Number of total unique genes in each ontology database.
go <- wppi_go_data() count_genes(go) # [1] 19712go <- wppi_go_data() count_genes(go) # [1] 19712
Filter ontology datasets using PPI network object
filter_annot_with_network(data_annot, graph_op)filter_annot_with_network(data_annot, graph_op)
data_annot |
Data frame (tibble) of GO or HPO datasets from
|
graph_op |
Igraph graph object obtained from built OmniPath PPI of genes of interest and x-degree neighbors. |
Data frame (tibble) of GO or HPO datasets filtered based on proteins available in the igraph object.
# Get GO database GO_data <- wppi_go_data() # Create igraph object based on genes of interest and first neighbors genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op <- graph_from_op(wppi_omnipath_data()) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter GO data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1)# Get GO database GO_data <- wppi_go_data() # Create igraph object based on genes of interest and first neighbors genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op <- graph_from_op(wppi_omnipath_data()) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter GO data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1)
Functional similarity between two genes in ontology databases (GO or HPO). Each pair of interacting proteins in the PPI graph network, is quantified the shared annotations between them using the Fisher's combined probability test (https://doi.org/10.1007/978-1-4612-4380-9_6). This is based on the number of genes annotated in each shared ontology term and the total amount of unique genes available in the ontology database.
functional_annot(annot, gene_i, gene_j)functional_annot(annot, gene_i, gene_j)
annot |
Processed annotation data as provided by
|
gene_i |
String with the gene symbol in the row of the adjacency matrix. |
gene_j |
String with the gene symbol in the column of the adjacency matrix. |
Numeric value with GO/HPO functional similarity between given pair of proteins.
hpo <- wppi_hpo_data() hpo <- process_annot(hpo) hpo_score <- functional_annot(hpo, 'AKT1', 'MTOR') # [1] 106.9376hpo <- wppi_hpo_data() hpo <- process_annot(hpo) hpo_score <- functional_annot(hpo, 'AKT1', 'MTOR') # [1] 106.9376
Creation of igraph object from PPI OmniPath database with information regarding proteins and gene symbols.
graph_from_op(op_data)graph_from_op(op_data)
op_data |
Data frame (tibble) of OmniPath PPI interactions from
|
Igraph PPI graph object with vertices defined by UniProt ID and Gene Symbol, and edges based on interactions, for all connections in OmniPath.
graph_op <- graph_from_op(wppi_omnipath_data()) edges_op <- igraph::E(graph_op) vertices_op <- igraph::V(graph_op)graph_op <- graph_from_op(wppi_omnipath_data()) edges_op <- igraph::E(graph_op) vertices_op <- igraph::V(graph_op)
Check which genes of interest are or not in OmniPath
in_omnipath(graph_op, gene_set, in_network = TRUE)in_omnipath(graph_op, gene_set, in_network = TRUE)
graph_op |
Igraph object based on OmniPath PPI interactions from
|
gene_set |
Character vector with known-disease specific genes from which is built the functional weighted PPI. |
in_network |
Logical: whether to return the genes in the network or the missing ones. |
Character vector with genes corresponding to the query.
# genes mapped and not mapped in OmniPath graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") genes_mapped <- in_omnipath(graph_op, genes_interest, 1) genes_notmapped <- in_omnipath(graph_op, genes_interest, 0)# genes mapped and not mapped in OmniPath graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") genes_mapped <- in_omnipath(graph_op, genes_interest, 1) genes_notmapped <- in_omnipath(graph_op, genes_interest, 0)
Ranks candidate genes based on correlation with the given seed genes of interest. For this, the source proteins/genes (i.e. starting nodes) are reduced to the candidate genes and the target proteins/genes (i.e. end nodes) to the given genes of interest. Each candidate gene score is defined by the sum of its correlations towards the known disease-related genes.
prioritization_genes( graph_op, prob_matrix, genes_interest, percentage_genes_ranked = 100 )prioritization_genes( graph_op, prob_matrix, genes_interest, percentage_genes_ranked = 100 )
graph_op |
Igraph object based on OmniPath PPI interactions from
|
prob_matrix |
Matrix object with correlations/probabilities of the
all nodes in the network from |
genes_interest |
Character vector with known-disease specific genes. |
percentage_genes_ranked |
Positive integer (range between 0 and 100) specifying the percentage ( network returned in the output. If not specified, the score of all the candidate genes is delivered. |
Data frame with the ranked candidate genes based on the functional score inferred from given ontology terms, PPI and Random Walk with Restart parameters.
db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered) # Random Walk with Restart w_rw <- random_walk(w_adj) # Ranked candidate genes scores <- prioritization_genes(graph_op_1, w_rw, genes_interest)db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered) # Random Walk with Restart w_rw <- random_walk(w_adj) # Ranked candidate genes scores <- prioritization_genes(graph_op_1, w_rw, genes_interest)
Ontology databases such as Gene Ontology (GO, http://geneontology.org/) and Human Phenotype Ontology (HPO, https://hpo.jax.org/app/) provide important genome and disease functional annotations of genes. These combined allow to build a connection between proteins/genes and phenotype/disease. This function aggregates information in the GO and HPO ontology datasets.
process_annot(data_annot)process_annot(data_annot)
data_annot |
Data frame (tibble) of GO or HPO datasets from
|
A list of four elements: 1) "term_size" a list which serves as a
lookup table for size (number of genes) for each ontology term; 2)
"gene_term" a list to look up terms by gene symbol; 3) "annot" the
original data frame (data_annot); 4) "total_genes" the number of
genes annotated in the ontology dataset.
hpo_raw <- wppi_hpo_data() hpo <- process_annot(hpo_raw)hpo_raw <- wppi_hpo_data() hpo <- process_annot(hpo_raw)
RWR on the normalized weighted adjacency matrix. The RWR algorithm estimates each protein/gene relevance based on the functional similarity of genes and disease/phenotype, and the topology of the network. This similarity score between nodes measures how closely two proteins/genes are related in a network. Thus, enabling to identify which candidate genes are more related to our given genes of interest.
random_walk(weighted_adj_matrix, restart_prob = 0.4, threshold = 1e-05)random_walk(weighted_adj_matrix, restart_prob = 0.4, threshold = 1e-05)
weighted_adj_matrix |
Matrix object corresponding to the weighted
adjacency from |
restart_prob |
Positive value between 0 and 1 defining the restart probability parameter used in the RWR algorithm. If not specified, 0.4 is the default value. |
threshold |
Positive value depicting the threshold parameter in the RWR algorithm. When the error between probabilities is smaller than the threshold defined, the algorithm stops. If not specified, 1e-5 is the default value. |
Matrix of correlation/probabilities for the functional similarities for all proteins/genes in the network.
db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered) # Random Walk with Restart w_rw <- random_walk(w_adj)db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered) # Random Walk with Restart w_rw <- random_walk(w_adj)
The wppi package implements a prioritization of genes according to their potential relevance in a disease or other experimental or physiological condition. For this it uses a PPI network and functional annotations. A protein-protein interactions (PPI) in the neighborhood of the genes of interest are weighted according to the number of common neighbors of interacting partners and the similarity of their functional annotations. The PPI networks are obtained using the OmniPath (https://omnipathdb.org/) resource and functionality is deduced using the Gene Ontology (GO, http://geneontology.org/) and Human Phenotype Ontology (HPO, https://hpo.jax.org/app/) ontology databases. To score the candidate genes, a Random Walk with Restart algorithm is applied on the weighted network.
score_candidate_genes_from_PPI( genes_interest, HPO_interest = NULL, percentage_output_genes = 100, graph_order = 1, GO_annot = TRUE, GO_slim = NULL, GO_aspects = c("C", "F", "P"), GO_organism = "human", HPO_annot = TRUE, restart_prob_rw = 0.4, threshold_rw = 1e-05, databases = NULL, ... )score_candidate_genes_from_PPI( genes_interest, HPO_interest = NULL, percentage_output_genes = 100, graph_order = 1, GO_annot = TRUE, GO_slim = NULL, GO_aspects = c("C", "F", "P"), GO_organism = "human", HPO_annot = TRUE, restart_prob_rw = 0.4, threshold_rw = 1e-05, databases = NULL, ... )
genes_interest |
Character vector of gene symbols with genes known to be related to the investigated disease or condition. |
HPO_interest |
Character vector with Human Phenotype Ontology (HPO)
annotations of interest from which to construct the functionality (for
a list of available annotations see the 'Name' column in the data
frame provided by |
percentage_output_genes |
Positive integer (range between 0 and 100) specifying the percentage (%) of the total candidate genes in the network returned in the output. If not specified, the score of all the candidate genes is delivered. |
graph_order |
Integer larger than zero: the neighborhood range counted as steps from the genes of interest. These genes, also called candidate genes, together with the given genes of interest define the Protein-Protein Interaction (PPI) network used in the analysis. If not specified, the first order neighbors are used. |
GO_annot |
Logical: use the Gene Ontology (GO) annotation database to weight the PPI network. The default is to use it. |
GO_slim |
Character: use a GO subset (slim). If |
GO_aspects |
Character vector with the single letter codes of the gene ontology aspects to use. By default all three aspects are used. The aspects are "C": cellular component, "F": molecular function and "P" biological process. |
GO_organism |
Character: name of the organism for GO annotations. |
HPO_annot |
Logical: use the Human Phenotype Ontology (HPO) annotation database to weight the PPI network. The default is to use it. |
restart_prob_rw |
Numeric: between 0 and 1, defines the restart probability parameter used in the Random Walk with Restart algorithm. The default value is 0.4. |
threshold_rw |
Numeric: the threshold parameter in the Random Walk with Restart algorithm. When the error between probabilities is smaller than the threshold, the algorithm stops. The default is 1e-5. |
databases |
Database knowledge as produced by |
... |
Passed to
|
If you use a GO subset (slim), building it at the first time might take
around 20 minutes. The result is saved into the cache so next time loading
the data from there is really quick.
Gene Ontology annotations are available for a few other organisms apart
from human. The currently supported organisms are "chicken", "cow", "dog",
"human", "pig" and "uniprot_all". If you disable HPO_annot you can
use wppi to score PPI networks other than human.
Data frame with the ranked candidate genes based on the functional score inferred from given ontology terms, PPI and Random Walk with Restart parameters.
# example gene set genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # example HPO annotations set hpo <- wppi_hpo_data() HPO_interest <- unique( dplyr::filter(hpo, grepl("Diabetes", .data$Name))$Name ) # Score 1st-order candidate genes new_genes_diabetes <- score_candidate_genes_from_PPI( genes_interest = genes_interest, HPO_interest = HPO_interest, percentage_output_genes = 10, graph_order = 1) new_genes_diabetes # # A tibble: 30 x 3 # score gene_symbol uniprot # <dbl> <chr> <chr> # 1 0.247 KNL1 Q8NG31 # 2 0.247 HTRA2 O43464 # 3 0.247 KAT6A Q92794 # 4 0.247 BABAM1 Q9NWV8 # 5 0.247 SKI P12755 # # . with 25 more rows# example gene set genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # example HPO annotations set hpo <- wppi_hpo_data() HPO_interest <- unique( dplyr::filter(hpo, grepl("Diabetes", .data$Name))$Name ) # Score 1st-order candidate genes new_genes_diabetes <- score_candidate_genes_from_PPI( genes_interest = genes_interest, HPO_interest = HPO_interest, percentage_output_genes = 10, graph_order = 1) new_genes_diabetes # # A tibble: 30 x 3 # score gene_symbol uniprot # <dbl> <chr> <chr> # 1 0.247 KNL1 Q8NG31 # 2 0.247 HTRA2 O43464 # 3 0.247 KAT6A Q92794 # 4 0.247 BABAM1 Q9NWV8 # 5 0.247 SKI P12755 # # . with 25 more rows
From the igraph object of a PPI network obtained from OmniPath extracts a
subnetwork around the provided genes of interest. The size of the graph
is determined by the sub_level parameter, i.e. the maximum number
of steps (order) from the genes of interest.
subgraph_op(graph_op, gene_set, sub_level = 1L)subgraph_op(graph_op, gene_set, sub_level = 1L)
graph_op |
Igraph object based on OmniPath PPI interactions from
|
gene_set |
Character vector of gene symbols. These are the genes of interest, for example known disease specific genes. |
sub_level |
Integer larger than 0 defining the order of neighborhood (number of steps) from the genes of interest. If not specified, the first-order neighbors are used. |
Igraph graph object with PPI network of given genes of interest and their x-order degree neighbors.
# Subgraphs of first and second order graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) graph_op_2 <- subgraph_op(graph_op, genes_interest, 2)# Subgraphs of first and second order graph_op <- graph_from_op(wppi_omnipath_data()) genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) graph_op_2 <- subgraph_op(graph_op, genes_interest, 2)
Converts adjacency to weighted adjacency using network topology
information (shared neighbors between connected nodes via
common_neighbors) integrated with genome and phenotype
factors from GO and HPO annotation terms (functionality computed by
functional_annot). At the end, the weighted adjacency
matrix is normalized by column.
weighted_adj(graph_op, GO_data, HPO_data)weighted_adj(graph_op, GO_data, HPO_data)
graph_op |
Igraph object based on OmniPath PPI interactions from
|
GO_data |
Data frame with GO annotations as provided by
|
HPO_data |
Data frame with HPO annotations as provided by
|
Weighted adjacency matrix based on network topology and functional similarity between interacting proteins/genes based on ontology databases.
db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered)db <- wppi_data() GO_data <- db$go HPO_data <- db$hpo # Genes of interest genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") # Graph object with PPI graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) # Filter ontology data GO_data_filtered <- filter_annot_with_network(GO_data, graph_op_1) HPO_data_filtered <- filter_annot_with_network(HPO_data, graph_op_1) # Weighted adjacency w_adj <- weighted_adj(graph_op_1, GO_data_filtered, HPO_data_filtered)
The wppi package calculates context specific scores for genes in
the network neighborhood of genes of interest. The context specificity
is ensured by the selection of the genes of interest and potentially by
using a more relevant subset of the ontology annotations, e.g. selecting
only the diabetes related categories. The PPI network and the functional
annotations are obtained automatically from public databases, though
it's possible to use custom databases. The network is limited to a
neighborhood of the genes of interest. The ontology annotations are also
filtered to the genes in this subnetwork. Then the adjacency matrix is
weighted according to the number of common neighbors and the similarity
in functional annotations of each pair of interacting proteins. On this
weighted adjacency matrix a random walk with restart is performed. The
final score for the genes in the neighborhood is the sum of their scores
(probabilities to be visited) in the random walk.
The method can be fine tuned by setting the neighborhood range, the
restart probability of the random walk and the threshold for the random
walk.
Ana Galhoz [email protected] and Denes Turei [email protected]
# Example with a single call: genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") scores <- score_candidate_genes_from_PPI(genes_interest) # The workflow step by step: db <- wppi_data() genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) w_adj <- weighted_adj(graph_op_1, db$go, db$hpo) w_rw <- random_walk(w_adj) scores <- prioritization_genes(graph_op_1, w_rw, genes_interest)# Example with a single call: genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") scores <- score_candidate_genes_from_PPI(genes_interest) # The workflow step by step: db <- wppi_data() genes_interest <- c("ERCC8", "AKT3", "NOL3", "GFI1B", "CDC25A", "TPX2", "SHE") graph_op <- graph_from_op(db$omnipath) graph_op_1 <- subgraph_op(graph_op, genes_interest, 1) w_adj <- weighted_adj(graph_op_1, db$go, db$hpo) w_rw <- random_walk(w_adj) scores <- prioritization_genes(graph_op_1, w_rw, genes_interest)
Retrieves the database knowledge necessary for WPPI directly from the databases. The databases used here are the Human Phenotype Ontology (HPO, https://hpo.jax.org/app/), Gene Ontology (GO, http://geneontology.org/) and OmniPath (https://omnipathdb.org/). The downloads carried out by the OmnipathR package and data required by wppi are extracted from each table.
wppi_data( GO_slim = NULL, GO_aspects = c("C", "F", "P"), GO_organism = "human", ... )wppi_data( GO_slim = NULL, GO_aspects = c("C", "F", "P"), GO_organism = "human", ... )
GO_slim |
Character: use a GO subset (slim). If |
GO_aspects |
Character vector with the single letter codes of the gene ontology aspects to use. By default all three aspects are used. The aspects are "C": cellular component, "F": molecular function and "P" biological process. |
GO_organism |
Character: name of the organism for GO annotations. |
... |
Passed to
|
If you use a GO subset (slim), building it at the first time might take
around 20 minutes. The result is saved into the cache so next time loading
the data from there is really quick.
Gene Ontology annotations are available for a few other organisms apart
from human. The currently supported organisms are "chicken", "cow", "dog",
"human", "pig" and "uniprot_all". If you disable HPO_annot you can
use wppi to score PPI networks other than human.
A list of data frames (tibbles) with database knowledge from HPO, GO and OmniPath.
# Download all data data_wppi <- wppi_data() # OmniPath omnipath_data <- data_wppi$omnipath # HPO HPO_data <- data_wppi$hpo # GO GO_data <- data_wppi$go# Download all data data_wppi <- wppi_data() # OmniPath omnipath_data <- data_wppi$omnipath # HPO HPO_data <- data_wppi$hpo # GO GO_data <- data_wppi$go
Gene Ontology (http://geneontology.org/), GO) annotates genes by their function, localization and biological processes.
wppi_go_data(slim = NULL, aspects = c("C", "F", "P"), organism = "human")wppi_go_data(slim = NULL, aspects = c("C", "F", "P"), organism = "human")
slim |
Character: use a GO subset (slim). If |
aspects |
Character vector with the single letter codes of the gene ontology aspects to use. By default all three aspects are used. The aspects are "C": cellular component, "F": molecular function and "P" biological process. |
organism |
Character: name of the organism for GO annotations. |
If you use a GO subset (slim), building it at the first time might take
around 20 minutes. The result is saved into the cache so next time loading
the data from there is really quick.
Gene Ontology annotations are available for a few other organisms apart
from human. The currently supported organisms are "chicken", "cow", "dog",
"human", "pig" and "uniprot_all". If you disable HPO_annot you can
use wppi to score PPI networks other than human.
A data frame (tibble) with GO annotation data.
go <- wppi_go_data()go <- wppi_go_data()
Human Phenotype Ontology (https://hpo.jax.org/app/), HPO) annotates proteins with phenotypes and diseases.
wppi_hpo_data()wppi_hpo_data()
A data frame (tibble) with HPO data.
hpo <- wppi_hpo_data()hpo <- wppi_hpo_data()
OmniPath (https://omnipathdb.org/) integrates protein-protein
interactions (PPI) from more than 30 resources. The network created is
highly customizable by passing parameters
to OmnipathR::import_post_translational_interactions.
wppi_omnipath_data(...)wppi_omnipath_data(...)
... |
Passed to
|
A data frame (tibble) with protein-protein interaction data from OmniPath.
omnipath <- wppi_omnipath_data()omnipath <- wppi_omnipath_data()