Package contains high-level functions for retrieving molecular interaction data from most public databases using PSICQUIC service (http://www.ebi.ac.uk/Tools/webservices/psicquic/view/main.xhtml) and manipulating that data. Useful features include: retrieve full interactome of a given species, retrieve all interactions between two taxonomy groups (taxonomy structure-aware search), query molecular interaction databases using MIQL, subset interaction data by list of protein/gene identifiers, interaction detection method, publication (PMIDs). Depends on PSICQUIC, data.table.
# Install R PItools package
install.packages("BiocManager") # for installing BioConductor dependencies
BiocManager::install("Biostrings", "remotes") # dependency for PSICQUIC & for installing from github
BiocManager::install("vitkl/PItools", dependencies = T)Load all interactions between human proteins in IntAct database.
Follow this example for more details.
# load package
library(PItools)
# load all interactions from IntAct FTP storage (https://www.ebi.ac.uk/intact/downloads)
human = fullInteractome(taxid = 9606, database = "IntActFTP",
format = "tab27",
clean = TRUE, # parse into usable format (takes 5-10 minutes)
protein_only = TRUE, # filter protein interactions
directory = NULL) # keep data files inside R library,
# or specify your directory # protein interactions lack directionality & are detected in multiple studies
# so find the number of unique interactions
uniqueNinteractions(human)
uniqueNinteractors(human)# subset by a list of proteins
mediator_complex_proteins = fread("https://www.uniprot.org/uniprot/?query=GO:0016592%20AND%20taxonomy:9606&format=tab&columns=id")
mediator_complex = subsetMITABbyID(human, ID_seed = mediator_complex_proteins$Entry,
within_seed = T, only_seed2nonseed = F)
uniqueNinteractions(mediator_complex)
uniqueNinteractors(mediator_complex)Load interactions for one protein identified with a specific method.
# Query for interactions of bacterial RNA polymerase sigma factor SigA
# identified using two-hybrid methods inall imex databases
# (result identical to IntActFTP but only requested interactions are downloaded)
inter = queryPSICQUICrlib(query = "id:P74565 AND detmethod:\"MI:0018\"",
format = "tab25",
database = "imex",
directory = "./data_files/")# The data returned by queryPSICQUICrlib constains auxillary information
# that is not necessary for most analysis. Let's clean the data.
inter = cleanMITAB(inter)
inter$data[1:5]Beware of study bias (ascertainment bias): in aggregated data better studied proteins have more interactions (not a problem for unbiased proteome-wide interaction screens).