Martin Garrido Rodriguez-Cordoba 2020-06-09
The main goal of this parser is to transform an Omnipath formatted set of interactions into an object usable by the mechanistic modelling tool hipathia. Hipathia uses a signal propagation algorithm to estimate the activity of receptor-to-effector signaling circuits. Its current version uses a pathway-centric approach fed by a mixture of context-specific and generic signaling pathways as defined by the KEGG classification. On the other hand, Omnipath is a meta-resource that contains biological information from different databases in a ready-to-use tabular format. It comprises several types of relationships between biological entities as protein-protein interactions or TF-target relationships, as well as gene and protein functional annotations.
The possible parser features derive from the curated information on Omnipath. Some examples:
-
Restrict to interactions from certain resources (i.e Reactome, BioGRID… etc).
-
Use the curation effort to create a range of results depending on the network reliability.
-
Expand the effector nodes to new effectors, using Kinase-Substrate or TF-target interactions.
-
Create a meta graph without filtering nodes by biological context.
Disclaimer: This repository includes the neccesary functions to perform the conversion of Omnipath interactions into hipathia MGI objects. The creation of the MGI object takes a really long time on complex graphs, specially when there are lots of edges between nodes which are not receptors nor effectors.
Load required packages
# main packages
library(OmnipathR)
library(hipathia)
# annotation package to translate symbols into entrezs
library(org.Hs.eg.db)
# packages to handle and plot graphs
library(igraph)
library(ggplot2)
library(ggraph)
library(Rgraphviz)
library(cowplot)
library(ggVennDiagram)
# other packages from tidyverse
library(dplyr)
library(purrr)
library(tidyr)
library(tibble)
# custom parsing and plotting funs
source("R/omnipath-to-hipathia.R")
source("R/visualize-hipathia-graph.R")
# set ggplot default theme
theme_set(theme_bw() + theme(plot.title = element_text(hjust = 0.5)))Import all Omnipath interactions with import_Omnipath_Interactions(),
which contains only interactions with references. Subset such
interactions to consensus stimulation or inhibition and discard
auto-interactions. The curation_effort attribute contains the the
number of unique resource+reference pairs per interaction.
interactions <- OmnipathR::import_Omnipath_Interactions()## Downloaded 36684 interactions
## removed 0 interactions during database filtering.
# filter only to directed interactions (consensus) and remove self interactions
intInteractions <- subset(interactions,
(consensus_stimulation == 1 & consensus_inhibition == 0) |
(consensus_stimulation == 0 & consensus_inhibition == 1)) %>%
subset(source_genesymbol != target_genesymbol)Explore number of interactions per resource
strsplit(intInteractions$sources, ";") %>%
unlist() %>%
data.frame(source = .) %>%
count(source) %>%
arrange(n) %>%
mutate(source = factor(source, levels = source)) %>%
ggplot(aes(x = n, y = source)) +
geom_bar(stat = "identity")
For this case of use, we will restrict interactions to those included in Reactome and BioGRID
selectedSources <- c("Reactome", "BioGRID")
intInteractions <- subset(intInteractions, grepl(pattern = paste0(selectedSources, collapse = "|"), x = sources))Hipathia uses a pathway-centric approach, which isolates nodes and interactions in different biological contexts defined by KEGG. It decomposes the signaling networks into a meta graph information object (MGI), which contains the main graph, the decomposed signaling circuits (in form of igraph objects) and the neccesary metadata to carry out the analysis. For this case of use, we will focus on the cell cycle pathway.
intPathways <- c("Cell Cycle"="hsa04110")
hipathiaMgi <- hipathia::load_pathways(species = "hsa", pathways_list = intPathways)## snapshotDate(): 2020-04-27
## Loaded 1 pathways
This is the complete network already processed to be used by Hipathia
beautyHipathiaGraph(hipathiaMgi$pathigraphs$hsa04110$graph)
Some of the decomposed subgraphs (all receptors that reach one effector)…
lapply(hipathiaMgi$pathigraphs$hsa04110$effector.subgraphs[1:2], beautyHipathiaGraph)## $`P-hsa04110-46`
##
## $`P-hsa04110-96 97`
For a first try, we will employ the subset of nodes from the cell cycle pathway, using Omnipath interactions to link the nodes. The first step consists on obtaining the list of genes (nodes) from hipathia:
intGenes <- V(hipathiaMgi$pathigraphs$hsa04110$graph)$genesList %>%
unlist() %>%
mapIds(x = org.Hs.eg.db, keys = ., keytype = "ENTREZID", column = "SYMBOL") %>%
as.character() %>%
.[!is.na(.)]## 'select()' returned 1:1 mapping between keys and columns
And subset the Omnipath interactions to those which contain a relation between genes in the pathway including complexes.
sourceIndex <- strsplit(intInteractions$source_genesymbol, "_") %>%
sapply(function(x) any(x %in% intGenes))
targetIndex <- strsplit(intInteractions$target_genesymbol, "_") %>%
sapply(function(x) any(x %in% intGenes))
# filter to interactions fulfilling the criteria
filteredInteractions <- intInteractions[sourceIndex & targetIndex, ]Once with the selected interactions, we can apply the omnipathToHipathia() function to transform this network into the hipathia MGI object. This function relies on the hipathia function mgi_from_sif(), which creates the MGI object from the sif and attribute files. The omnipathToHipathia() will format the network and write the files into a temporary location in order to be imported with hipathia.
omnipathMgi <- omnipathToHipathia(filteredInteractions)## 'select()' returned 1:1 mapping between keys and columns
## `summarise()` ungrouping output (override with `.groups` argument)
## Parsing a graph with 61 nodes and 88 edges.
## Writing on /tmp/Rtmpee236y
## Starting hipathia mgi creation from sif...
## Loading graphs...
## Creating MGI...
## Created MGI with 1 pathway(s)
## Done!
We can now perform a visual comparison of the networks
default <- beautyHipathiaGraph(g = hipathiaMgi$pathigraphs$hsa04110$graph) + ggtitle("Default Cell Cycle")
op <- beautyHipathiaGraph(g = omnipathMgi$pathigraphs$hsa00$graph) + ggtitle("Omnipath derived Cell Cycle")## Warning in .local(from, to, graph): edges replaced: 'N-hsa00-25|N-hsa00-21'
## Warning in .local(from, to, graph): edges replaced: 'N-hsa00-25|N-hsa00-43'
## Warning in .local(from, to, graph): edges replaced: 'N-hsa00-25|N-hsa00-20'
cowplot::plot_grid(default, op)
And explore how many gene-to-gene interactions are shared between both MGIs
# create a graph list
graphs <- list(default = hipathiaMgi$pathigraphs$hsa04110$graph, omnipath = omnipathMgi$pathigraphs$hsa00$graph)
# get all possible gene to gene interactions
geneToGene <- lapply(graphs, function(g) {
dfs <- igraph::as_data_frame(g, what = "both")
isNull <- sapply(dfs$vertices$genesList, is.null)
dfs$vertices$genesList[isNull] <- NA
geneToGene <- tibble(from = dfs$vertices[dfs$edges$from, "genesList"], to = dfs$vertices[dfs$edges$to, "genesList"]) %>%
tidyr::unnest(from) %>%
tidyr::unnest(to) %>%
subset(from != "/" & !is.na(from) & to != "/" & !is.na(to)) %>%
transmute(id = paste0(from, "-->", to)) %>%
pull(id)
return(geneToGene)
})
# explore overlap between two sets
ggVennDiagram::ggVennDiagram(geneToGene)
The omnipathMGI object is now ready to be passed to functions from
hipathia such as:
hipathia()normalize_paths()get_path_names()
sessionInfo()## R version 4.0.1 (2020-06-06)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 18.04.4 LTS
##
## Matrix products: default
## BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.7.1
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.7.1
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=es_ES.UTF-8 LC_COLLATE=en_US.UTF-8
## [5] LC_MONETARY=es_ES.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=es_ES.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=es_ES.UTF-8 LC_IDENTIFICATION=C
##
## attached base packages:
## [1] grid stats4 parallel stats graphics grDevices utils
## [8] datasets methods base
##
## other attached packages:
## [1] tibble_3.0.1 tidyr_1.1.0
## [3] purrr_0.3.4 dplyr_1.0.0
## [5] ggVennDiagram_0.3 cowplot_1.0.0
## [7] Rgraphviz_2.32.0 graph_1.66.0
## [9] ggraph_2.0.3 ggplot2_3.3.1
## [11] org.Hs.eg.db_3.11.4 AnnotationDbi_1.50.0
## [13] hipathia_2.4.0 MultiAssayExperiment_1.14.0
## [15] SummarizedExperiment_1.18.1 DelayedArray_0.14.0
## [17] matrixStats_0.56.0 Biobase_2.48.0
## [19] GenomicRanges_1.40.0 GenomeInfoDb_1.24.0
## [21] IRanges_2.22.2 S4Vectors_0.26.1
## [23] AnnotationHub_2.20.0 BiocFileCache_1.12.0
## [25] dbplyr_1.4.4 BiocGenerics_0.34.0
## [27] OmnipathR_1.2.0 igraph_1.2.5
##
## loaded via a namespace (and not attached):
## [1] bitops_1.0-6 sf_0.9-3
## [3] bit64_0.9-7 httr_1.4.1
## [5] tools_4.0.1 R6_2.4.1
## [7] KernSmooth_2.23-17 DBI_1.1.0
## [9] colorspace_1.4-1 withr_2.2.0
## [11] tidyselect_1.1.0 gridExtra_2.3
## [13] VennDiagram_1.6.20 bit_1.1-15.2
## [15] curl_4.3 compiler_4.0.1
## [17] preprocessCore_1.50.0 formatR_1.7
## [19] labeling_0.3 scales_1.1.1
## [21] classInt_0.4-3 rappdirs_0.3.1
## [23] stringr_1.4.0 digest_0.6.25
## [25] rmarkdown_2.2 XVector_0.28.0
## [27] pkgconfig_2.0.3 htmltools_0.4.0
## [29] fastmap_1.0.1 rlang_0.4.6
## [31] RSQLite_2.2.0 shiny_1.4.0.2
## [33] generics_0.0.2 farver_2.0.3
## [35] jsonlite_1.6.1 RCurl_1.98-1.2
## [37] magrittr_1.5 GenomeInfoDbData_1.2.3
## [39] futile.logger_1.4.3 Matrix_1.2-18
## [41] Rcpp_1.0.4.6 munsell_0.5.0
## [43] viridis_0.5.1 lifecycle_0.2.0
## [45] stringi_1.4.6 yaml_2.2.1
## [47] MASS_7.3-51.6 zlibbioc_1.34.0
## [49] blob_1.2.1 promises_1.1.0
## [51] ggrepel_0.8.2 crayon_1.3.4
## [53] lattice_0.20-41 graphlayouts_0.7.0
## [55] knitr_1.28 pillar_1.4.4
## [57] futile.options_1.0.1 servr_0.17
## [59] glue_1.4.1 BiocVersion_3.11.1
## [61] evaluate_0.14 lambda.r_1.2.4
## [63] BiocManager_1.30.10 vctrs_0.3.1
## [65] tweenr_1.0.1 httpuv_1.5.4
## [67] gtable_0.3.0 polyclip_1.10-0
## [69] assertthat_0.2.1 xfun_0.14
## [71] ggforce_0.3.1 mime_0.9
## [73] xtable_1.8-4 tidygraph_1.2.0
## [75] e1071_1.7-3 later_1.1.0.1
## [77] class_7.3-17 viridisLite_0.3.0
## [79] memoise_1.1.0 units_0.6-6
## [81] ellipsis_0.3.1 interactiveDisplayBase_1.26.2
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent