MOSClip Two-Class Analysis on TCGA ovarian cancer patients
Prepare data for MOSClip analysis
Now we are almost ready to run a two-class analysis with MOSClip! First of all, we can load necessary libraries and the pre-processed ovarian cancer datasets that we downloaded from TCGA in the previous tutorials. We will also set a seed, in order to have reproducible results in case of future repetition of the analysis, and we will create a directory where we will save all the results generated by this tutorial.
library(org.Hs.eg.db)
library(EDASeq)
library(graphite)
library(MOSClip)
library(kableExtra)
load("downloadTCGA/TCGA-OV-pre-processed.RData")
set.seed(1234)
dirname <- "MOSresults/twoClass/"
if (!dir.exists(dirname)){
dir.create(dirname)
}We need to prepare data in order to run MOSClip. The first step is to
modify all the multi-omic matrices assigning the type of gene
identifier. Since we will use graphite to download a list
of pathways and their graphical structure, we need to format gene names
for each omic in order to be compatible with graphite.
Here, we will work with Entrez Gene ID, thus we have to indicate with
the prefix tag “ENTREZID:” each gene.
expression <- expAvg
row.names(expression) <- paste0("ENTREZID:", row.names(expression))
mutation <- ov.mutations$data
row.names(mutation) <- paste0("ENTREZID:", row.names(mutation))
names(metClustValues$eMap) <- paste0("ENTREZID:", row.names(metClustValues$eMap))
row.names(ov.cnv) <- paste0("ENTREZID:", row.names(ov.cnv))Moving to patient selection, we will keep only patients whose class annotation is available and their intersection with available patients across the 4 omics. Finally, our class annotation dataframe will be filtered to keep only selected patients.
# select common patients
patients <- row.names(classes)
patients <- intersect(patients, colnames(expression))
patients <- intersect(patients, colnames(metClustValues$MET_Cancer_Clustered))
patients <- intersect(patients, colnames(mutation))
patients <- intersect(patients, colnames(ov.cnv))
classAnnot <- classes[patients, , drop=FALSE]
table(classAnnot)## classes
## Differentiated Immunoreactive Mesenchymal Proliferative
## 74 67 64 63At this point, we need to extract selected patients for each
multi-omic matrix. After patient selection, we can normalize (upper
quartile normalization from EDASeq package) and
log-transform expression data.
# normalize expression data
expression <- expression[, patients, drop = FALSE]
keep = apply(expression >= 100, 1, any)
expNorm <- betweenLaneNormalization(expression[keep, , drop = FALSE], which = "upper")
pseudoExpNorm <- log2(expNorm + 1)
methylation <- metClustValues
methylation$MET_Cancer_Clustered <- methylation$MET_Cancer_Clustered[, patients, drop = FALSE]
mutation <- mutation[, patients, drop = FALSE]
cnv <- ov.cnv[, patients, drop = FALSE]We are now ready to generate an object of class Omics
using MOSClip function makeOmics. This object is required
to run each type of MOSClip analysis. It is based on
MultiAssayExperiment object, containing an
ExperimentList with matrices for each omic (we suggest to
use standard names for each experiment as shown in this example).
colData in this case will contain class annotation for
patients. Additionally, specific slots for MOSClip analysis
exist, including modelInfo, where the user must specify the
desired method for data reduction for each omic, and
specificArgs, with specific parameters to be used by
reduction functions. Both these slots must have the same dimension as
ExperimentList.
The list of available methods for data dimensionality reduction is
easily visualized with availableOmicsMethods().
multiOmics <- makeOmics(experiments = list(exp = pseudoExpNorm,
met = methylation$MET_Cancer_Clustered,
mut = mutation,
cnv = as.matrix(cnv)),
colData = classAnnot,
modelInfo = c("summarizeWithPca", "summarizeInCluster",
"summarizeToNumberOfEvents", "summarizeToNumberOfDirectionalEvents"),
specificArgs = list(pcaArgs = list(name = "exp", shrink = "FALSE", method = "sparse", maxPCs = 3),
clusterArgs = list(name = "met", dict = methylation$eMap, max_cluster_number = 3),
countEvent = list(name = "mut", min_prop = 0.05),
cnvAgv = list(name = "cnv", min_prop = 0.05)))Download Reactome pathways
To run a MOSClip analysis we also need the list of
pathways that we want to test, as well as their graphical structure if
we want to exploit the topological analysis.
We decide to use pathways collected in Reactome database. They can be
downloaded using graphite that is also able to convert gene
identifiers (here we will use Entrez Gene ID). This process may take
some minutes, for this reason we will save the final
PathwayList object, ready to be used in future
analyses.
if (file.exists("downloadTCGA/reactome-entrez-2024-05-27.RData")) {
load("downloadTCGA/reactome-entrez-2024-05-27.RData")
} else {
reactome <- pathways("hsapiens", "reactome")
reactome <- convertIdentifiers(reactome, "entrez")
file = paste0("downloadTCGA", "/reactome-entrez-", as.character(Sys.Date()), ".RData")
save(reactome, file = file)
}To avoid redundant and too heavy analyses, we select Reactome
pathways based on their number of nodes, considering only nodes that are
present at least in the expression matrix. We prepare two distinct
PathwayList objects:
- one containing pathways with more than 20 and less than 100
nodes,
- a larger one with pathways having more than 10 and less than 700 nodes.
nodesLength <- sapply(reactome, function(g) {length(intersect(graphite::nodes(g),
row.names(pseudoExpNorm)))})
reactSmall <- reactome[nodesLength >= 20 & nodesLength <= 100]
reactHuge <- reactome[nodesLength >= 10 & nodesLength <= 700]Prepare class annotations
We define the classes that we want to compare. Here we are using the subtypes defined for TCGA ovarian cancer patients. For the purpose of this tutorial, it could be interesting to compare on a multi-omic level immunoreactive patients with mesenchymal patients. The user can decide which subtypes to compare.
We can filter accordingly our class annotation data frame and
multiOmics object.
class1 <- "Immunoreactive"
class2 <- "Mesenchymal"
classAnnotation <- classAnnot[classAnnot$classes %in% c(class1, class2), , drop=FALSE]
multiOmics <- multiOmics[,row.names(classAnnotation)]Now we are ready for MOSClip two-class analysis.
Two-class analysis on modules
We start from the analysis on modules using the function
multiOmicsTwoClassModuleTest. Required inputs are an
Omics object, a graph, and a class annotation data frame.
Patients in class annotation should have the same order as in
colData. Reactome database is built with a hierarchical
organization: different pathways represent the same process with more or
less details, going from very specific and small pathways to very large
and general ones. To avoid redundant results, we choose focus on a
subset of pathways, specifically those having between 20 and 100 nodes
(reactSmall list); this way, we aim to limit the overlap in
gene testing.
This step will take some minutes. For this reason, we will save the final output in the directory “MOSresults”, and in case of future analyses we will load it from there.
if (file.exists(paste0(dirname, "twoClassM.rds"))){
twoClassM <- readRDS(paste0(dirname, "twoClassM.rds"))
} else {
twoClassM <- lapply(reactSmall, function(g) {
res <- multiOmicsTwoClassModuleTest(multiOmics, g, classAnnotation,
useTheseGenes = row.names(pseudoExpNorm))
res
})
saveRDS(twoClassM, file = paste0(dirname, "twoClassM.rds"))
}The result is a list of MultiOmicsModules objects. This
list can be used within the function
multiPathwayModuleReport to return a tabular summary of
modules, sorted by p-values. Besides the p-value for each module,
p-values for tested covariates are shown. The user can specify which
covariates to visualize first in the data frame columns, giving the omic
names to priority_to argument.
moduleSummary <- multiPathwayModuleReport(twoClassM,
priority_to = c("exp", "met", "cnv"))| module | pvalue | expPC1 | expPC2 | expPC3 | met2k2 | met3k2 | met3k3 | cnvNEG | cnvPOS | mut | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Elastic fibre formation.5 | 5 | 1.0e-07 | 0.0000001 | 0.0430093 | NA | 0.5044956 | NA | NA | NA | 0.5472875 | NA |
| SUMOylation of transcription factors.4 | 4 | 1.0e-07 | 0.0000000 | NA | NA | NA | NA | NA | NA | 0.5666869 | NA |
| Elastic fibre formation.2 | 2 | 1.0e-07 | 0.0000000 | NA | NA | 0.4479830 | NA | NA | NA | NA | NA |
| O-glycosylation of TSR domain-containing proteins.4 | 4 | 2.0e-07 | 0.0000000 | NA | NA | 0.9925894 | NA | NA | NA | 0.2557571 | NA |
| Gamma carboxylation, hypusinylation, hydroxylation, and arylsulfatase activation.4 | 4 | 2.0e-07 | 0.0000000 | NA | NA | 0.6992146 | NA | NA | NA | 0.0453129 | NA |
| Glycosphingolipid metabolism.11 | 11 | 2.0e-07 | 0.0000000 | NA | NA | 0.6992146 | NA | NA | NA | 0.0453129 | NA |
| Sphingolipid metabolism.15 | 15 | 2.0e-07 | 0.0000000 | NA | NA | 0.6992146 | NA | NA | NA | 0.0453129 | NA |
| Glycosphingolipid catabolism.8 | 8 | 2.0e-07 | 0.0000000 | NA | NA | 0.6992146 | NA | NA | NA | 0.0453129 | NA |
| ECM proteoglycans.2 | 2 | 3.0e-07 | 0.1654198 | 0.1413438 | 0.0000425 | 0.0845938 | NA | NA | NA | 0.4293656 | NA |
| RUNX2 regulates osteoblast differentiation.4 | 4 | 3.0e-07 | 0.0000001 | NA | NA | NA | NA | NA | 0.8990975 | 0.6534472 | NA |
| RUNX2 regulates bone development.5 | 5 | 3.0e-07 | 0.0000001 | NA | NA | NA | NA | NA | 0.8990975 | 0.6534472 | NA |
| Nitric oxide stimulates guanylate cyclase.2 | 2 | 3.0e-07 | 0.2998266 | 0.0000000 | NA | NA | NA | NA | NA | 0.0574532 | NA |
| O-glycosylation of TSR domain-containing proteins.23 | 23 | 3.0e-07 | 0.0000000 | NA | NA | 0.2906849 | NA | NA | NA | NA | NA |
| ECM proteoglycans.1 | 1 | 4.0e-07 | 0.0241493 | 0.0093753 | 0.0000246 | NA | 0.5571007 | 0.3716193 | NA | 0.3592535 | NA |
| Assembly of collagen fibrils and other multimeric structures.1 | 1 | 5.0e-07 | 0.0000002 | NA | NA | 0.0526327 | NA | NA | NA | 0.3959735 | NA |
| Collagen chain trimerization.1 | 1 | 5.0e-07 | 0.0000002 | NA | NA | 0.0526327 | NA | NA | NA | 0.3959735 | NA |
| Chondroitin sulfate biosynthesis.2 | 2 | 7.0e-07 | 0.9921860 | 0.0000002 | 0.3845546 | 0.4312105 | NA | NA | 0.6242335 | 0.1703916 | NA |
| Bacterial Infection Pathways.1 | 1 | 8.0e-07 | 0.0000000 | NA | NA | NA | 0.5747115 | 0.9897028 | NA | NA | NA |
| Molecules associated with elastic fibres.1 | 1 | 8.0e-07 | 0.0000231 | 0.0519096 | 0.0007252 | 0.3218417 | NA | NA | 0.7926718 | 0.1662188 | NA |
| Non-integrin membrane-ECM interactions.4 | 4 | 8.0e-07 | 0.0000004 | 0.0000123 | NA | NA | 0.4979362 | 0.9528540 | NA | 0.7731634 | NA |
| Signaling by TGF-beta Receptor Complex.12 | 12 | 9.0e-07 | 0.0000000 | NA | NA | 0.0370336 | NA | NA | 0.6218599 | 0.0552526 | NA |
| Transcriptional activity of SMAD2/SMAD3:SMAD4 heterotrimer.5 | 5 | 9.0e-07 | 0.0000000 | NA | NA | 0.0370336 | NA | NA | 0.6218599 | 0.0552526 | NA |
| SMAD2/SMAD3:SMAD4 heterotrimer regulates transcription.4 | 4 | 9.0e-07 | 0.0000000 | NA | NA | 0.0370336 | NA | NA | 0.6218599 | 0.0552526 | NA |
| Assembly of collagen fibrils and other multimeric structures.4 | 4 | 9.0e-07 | 0.0000164 | 0.7065204 | NA | 0.0888073 | NA | NA | NA | 0.1326165 | 0.7739622 |
| Collagen chain trimerization.4 | 4 | 9.0e-07 | 0.0000164 | 0.7065204 | NA | 0.0888073 | NA | NA | NA | 0.1326165 | 0.7739622 |
| Unfolded Protein Response (UPR).23 | 23 | 1.0e-06 | 0.0000004 | NA | NA | NA | 0.5221858 | 0.8471660 | NA | NA | NA |
| O-glycosylation of TSR domain-containing proteins.34 | 34 | 1.1e-06 | 0.0000000 | NA | NA | 0.8321512 | NA | NA | NA | 0.5098868 | NA |
| Signaling by ALK in cancer.10 | 10 | 1.2e-06 | 0.2136126 | 0.0000001 | 0.0254404 | 0.1086051 | NA | NA | NA | 0.5563069 | NA |
| Signaling by ALK fusions and activated point mutants.10 | 10 | 1.2e-06 | 0.2136126 | 0.0000001 | 0.0254404 | 0.1086051 | NA | NA | NA | 0.5563069 | NA |
| Cell junction organization.6 | 6 | 1.3e-06 | 0.0000000 | NA | NA | 0.6829094 | NA | NA | NA | 0.8181169 | NA |
| O-glycosylation of TSR domain-containing proteins.28 | 28 | 1.4e-06 | 0.0000000 | NA | NA | 0.7965990 | NA | NA | NA | 0.7398864 | NA |
| Chondroitin sulfate biosynthesis.3 | 3 | 1.4e-06 | 0.5245592 | 0.0000006 | 0.1625020 | NA | 0.5076500 | 0.5248538 | 0.3964371 | 0.3221811 | NA |
| Collagen formation.2 | 2 | 1.4e-06 | 0.0000010 | 0.0050478 | 0.7565458 | 0.1563999 | NA | NA | 0.3859780 | 0.3054711 | 0.8282875 |
| Immunoregulatory interactions between a Lymphoid and a non-Lymphoid cell.14 | 14 | 1.5e-06 | 0.0332499 | 0.0000000 | NA | 0.9115779 | NA | NA | NA | 0.8072930 | NA |
| Signaling by FGFR.10 | 10 | 1.6e-06 | 0.4196746 | 0.0000000 | 0.0020458 | 0.3525830 | NA | NA | 0.0782642 | 0.7629884 | NA |
| Signaling by FGFR1.8 | 8 | 1.6e-06 | 0.4196746 | 0.0000000 | 0.0020458 | 0.3525830 | NA | NA | 0.0782642 | 0.7629884 | NA |
| Collagen formation.3 | 3 | 1.7e-06 | 0.0000010 | 0.0043916 | 0.7007774 | 0.2173066 | NA | NA | 0.3931109 | 0.4241760 | 0.9316167 |
| Collagen biosynthesis and modifying enzymes.2 | 2 | 1.7e-06 | 0.0000010 | 0.0043916 | 0.7007774 | 0.2173066 | NA | NA | 0.3931109 | 0.4241760 | 0.9316167 |
| Activation of Matrix Metalloproteinases.2 | 2 | 1.7e-06 | 0.0000233 | 0.5657766 | 0.0000687 | 0.0710426 | NA | NA | NA | 0.6188975 | NA |
| XBP1(S) activates chaperone genes.25 | 25 | 1.8e-06 | 0.0000000 | NA | NA | NA | 0.1578483 | 0.4096951 | NA | 0.0716312 | NA |
| IRE1alpha activates chaperones.27 | 27 | 1.8e-06 | 0.0000000 | NA | NA | NA | 0.1578483 | 0.4096951 | NA | 0.0716312 | NA |
| Unfolded Protein Response (UPR).31 | 31 | 1.8e-06 | 0.0000000 | NA | NA | NA | 0.1578483 | 0.4096951 | NA | 0.0716312 | NA |
| Heparan sulfate/heparin (HS-GAG) metabolism.7 | 7 | 1.9e-06 | 0.0000001 | 0.0000002 | 0.5249690 | 0.2119964 | NA | NA | NA | 0.6004117 | NA |
| A tetrasaccharide linker sequence is required for GAG synthesis.12 | 12 | 1.9e-06 | 0.0000001 | 0.0000002 | 0.5249690 | 0.2119964 | NA | NA | NA | 0.6004117 | NA |
| Signaling by FGFR in disease.6 | 6 | 2.0e-06 | 0.2932271 | 0.0000001 | 0.0116695 | 0.0145332 | NA | NA | 0.8262102 | 0.7253705 | 0.1078824 |
| Signaling by FGFR2 in disease.1 | 1 | 2.0e-06 | 0.2932271 | 0.0000001 | 0.0116695 | 0.0145332 | NA | NA | 0.8262102 | 0.7253705 | 0.1078824 |
| MAPK6/MAPK4 signaling.6 | 6 | 2.0e-06 | 0.0000000 | NA | NA | NA | 0.0008409 | 0.1546065 | NA | 0.3455090 | NA |
| Interleukin-10 signaling.35 | 35 | 2.1e-06 | 0.0000000 | NA | NA | NA | NA | NA | NA | 0.3584628 | NA |
| Signaling by BMP.1 | 1 | 2.5e-06 | 0.0000001 | 0.9396654 | 0.3636854 | 0.3992086 | NA | NA | 0.8247345 | 0.0344677 | NA |
| Chondroitin sulfate/dermatan sulfate metabolism.7 | 7 | 2.5e-06 | 0.4790755 | 0.0000000 | 0.4828802 | 0.9397531 | NA | NA | 0.2800561 | 0.4183288 | 0.8148043 |
| Chondroitin sulfate biosynthesis.4 | 4 | 2.6e-06 | 0.3936016 | 0.7777561 | 0.0000001 | 0.0252156 | NA | NA | 0.0562418 | 0.6133604 | NA |
| Collagen formation.1 | 1 | 2.6e-06 | 0.0000007 | 0.0020706 | 0.6738027 | 0.1884023 | NA | NA | 0.0822127 | 0.1264153 | 0.8214458 |
| Collagen biosynthesis and modifying enzymes.1 | 1 | 2.6e-06 | 0.0000007 | 0.0020706 | 0.6738027 | 0.1884023 | NA | NA | 0.0822127 | 0.1264153 | 0.8214458 |
| Diseases associated with glycosaminoglycan metabolism.4 | 4 | 2.6e-06 | 0.0887587 | 0.2683040 | 0.0000001 | NA | NA | NA | 0.7528869 | 0.9907462 | 0.8913413 |
| Collagen formation.6 | 6 | 2.7e-06 | 0.0000117 | 0.4137637 | 0.0598106 | 0.8867219 | NA | NA | 0.5096098 | 0.3257258 | NA |
| Assembly of collagen fibrils and other multimeric structures.13 | 13 | 2.7e-06 | 0.0000117 | 0.4137637 | 0.0598106 | 0.8867219 | NA | NA | 0.5096098 | 0.3257258 | NA |
| Platelet homeostasis.2 | 2 | 2.7e-06 | 0.4200899 | 0.0000002 | 0.0070693 | 0.1039861 | NA | NA | NA | 0.3347172 | NA |
| Syndecan interactions.3 | 3 | 2.8e-06 | 0.0001821 | 0.0000002 | NA | NA | 0.0325411 | 0.3753025 | NA | 0.3075261 | NA |
| Non-integrin membrane-ECM interactions.5 | 5 | 2.8e-06 | 0.0001821 | 0.0000002 | NA | NA | 0.0325411 | 0.3753025 | NA | 0.3075261 | NA |
| Signaling by FGFR in disease.5 | 5 | 2.9e-06 | 0.3606012 | 0.0000004 | 0.0011555 | 0.2837853 | NA | NA | 0.9731297 | 0.9571113 | NA |
| Assembly of collagen fibrils and other multimeric structures.7 | 7 | 2.9e-06 | 0.0000000 | NA | NA | 0.9918623 | NA | NA | NA | NA | NA |
| Collagen chain trimerization.7 | 7 | 2.9e-06 | 0.0000000 | NA | NA | 0.9918623 | NA | NA | NA | NA | NA |
| O-glycosylation of TSR domain-containing proteins.31 | 31 | 2.9e-06 | 0.0000000 | NA | NA | 0.9906175 | NA | NA | NA | 0.0072528 | NA |
| Signaling by FGFR in disease.3 | 3 | 3.1e-06 | 0.3575306 | 0.0000004 | 0.0011815 | 0.2877414 | NA | NA | 0.9248441 | 0.9168705 | NA |
| Signaling by FGFR in disease.4 | 4 | 3.1e-06 | 0.3225506 | 0.0000001 | 0.0010588 | 0.3536394 | NA | NA | 0.9670913 | 0.7881869 | NA |
| Syndecan interactions.7 | 7 | 3.2e-06 | 0.0071035 | 0.8405068 | 0.0000001 | NA | 0.0099464 | 0.7105744 | 0.5835903 | 0.3219886 | NA |
| Non-integrin membrane-ECM interactions.9 | 9 | 3.2e-06 | 0.0071035 | 0.8405068 | 0.0000001 | NA | 0.0099464 | 0.7105744 | 0.5835903 | 0.3219886 | NA |
| Signaling by FGFR2 in disease.2 | 2 | 3.6e-06 | 0.1727524 | 0.0000001 | 0.0111432 | 0.0438023 | NA | NA | 0.4370921 | 0.8553761 | NA |
| ECM proteoglycans.8 | 8 | 3.6e-06 | 0.0000000 | NA | NA | 0.0888882 | NA | NA | 0.4096324 | 0.8958364 | 0.3549886 |
| Activation of Matrix Metalloproteinases.3 | 3 | 3.6e-06 | 0.0000001 | 0.1796756 | 0.3827438 | 0.0586231 | NA | NA | NA | 0.4440382 | NA |
| FRS-mediated FGFR2 signaling.1 | 1 | 3.6e-06 | 0.1566822 | 0.0000001 | 0.0118130 | 0.0447017 | NA | NA | 0.4599664 | 0.3875140 | NA |
| Diseases associated with glycosaminoglycan metabolism.1 | 1 | 3.6e-06 | 0.0000063 | 0.2435271 | 0.2465458 | 0.0203295 | NA | NA | 0.4735686 | 0.8517177 | 0.9448151 |
| Defective B4GALT7 causes EDS, progeroid type.1 | 1 | 3.6e-06 | 0.0000063 | 0.2435271 | 0.2465458 | 0.0203295 | NA | NA | 0.4735686 | 0.8517177 | 0.9448151 |
| FGFR2 mutant receptor activation.1 | 1 | 3.7e-06 | 0.1855737 | 0.0000001 | 0.0115202 | 0.1384827 | NA | NA | 0.7255324 | 0.5102733 | NA |
| Diseases associated with glycosaminoglycan metabolism.3 | 3 | 3.9e-06 | 0.0000070 | 0.2229971 | 0.2492377 | 0.0197037 | NA | NA | 0.4518418 | 0.9549673 | 0.9406503 |
| Defective B3GAT3 causes JDSSDHD.1 | 1 | 3.9e-06 | 0.0000070 | 0.2229971 | 0.2492377 | 0.0197037 | NA | NA | 0.4518418 | 0.9549673 | 0.9406503 |
| Diseases associated with glycosaminoglycan metabolism.10 | 10 | 3.9e-06 | 0.0000000 | NA | NA | 0.9883884 | NA | NA | NA | 0.2846400 | NA |
| Diseases associated with glycosaminoglycan metabolism.2 | 2 | 4.0e-06 | 0.0000077 | 0.2174316 | 0.2502998 | 0.0198502 | NA | NA | 0.4427936 | 0.8482583 | 0.9372121 |
| Defective B3GALT6 causes EDSP2 and SEMDJL1.1 | 1 | 4.0e-06 | 0.0000077 | 0.2174316 | 0.2502998 | 0.0198502 | NA | NA | 0.4427936 | 0.8482583 | 0.9372121 |
| SHC-mediated cascade:FGFR2.1 | 1 | 4.0e-06 | 0.1582481 | 0.0000001 | 0.0114429 | 0.0805351 | NA | NA | 0.4707031 | 0.5287020 | NA |
| Chondroitin sulfate biosynthesis.1 | 1 | 4.0e-06 | 0.4138422 | 0.0000000 | 0.9567405 | 0.1275922 | NA | NA | 0.1305718 | 0.2464359 | NA |
| Collagen formation.5 | 5 | 4.1e-06 | 0.0000241 | 0.1134608 | 0.7455172 | 0.5377312 | NA | NA | 0.2002979 | 0.3300364 | 0.1098756 |
| Assembly of collagen fibrils and other multimeric structures.12 | 12 | 4.1e-06 | 0.0000241 | 0.1134608 | 0.7455172 | 0.5377312 | NA | NA | 0.2002979 | 0.3300364 | 0.1098756 |
| O-glycosylation of TSR domain-containing proteins.1 | 1 | 4.1e-06 | 0.0000000 | NA | NA | 0.8003887 | NA | NA | NA | 0.9922676 | NA |
| COPI-dependent Golgi-to-ER retrograde traffic.1 | 1 | 4.1e-06 | 0.0734397 | 0.0000003 | 0.0172880 | NA | 0.0065869 | 0.2067050 | 0.2089244 | 0.2319105 | 0.4636113 |
| Binding and Uptake of Ligands by Scavenger Receptors.6 | 6 | 4.1e-06 | 0.0000098 | 0.4437830 | 0.0102223 | NA | 0.2350263 | 0.9940394 | 0.3271237 | 0.7874115 | 0.0133496 |
| Heparan sulfate/heparin (HS-GAG) metabolism.3 | 3 | 4.2e-06 | 0.0000000 | 0.0000704 | 0.8200092 | 0.3476560 | NA | NA | 0.2220251 | 0.8520836 | NA |
| A tetrasaccharide linker sequence is required for GAG synthesis.7 | 7 | 4.2e-06 | 0.0000000 | 0.0000704 | 0.8200092 | 0.3476560 | NA | NA | 0.2220251 | 0.8520836 | NA |
| Downstream signaling of activated FGFR2.2 | 2 | 4.2e-06 | 0.1429566 | 0.0000001 | 0.0115034 | 0.0836176 | NA | NA | 0.4721888 | 0.2036482 | NA |
| Signaling by FGFR2.3 | 3 | 4.2e-06 | 0.1429566 | 0.0000001 | 0.0115034 | 0.0836176 | NA | NA | 0.4721888 | 0.2036482 | NA |
| Negative regulation of FGFR2 signaling.1 | 1 | 4.2e-06 | 0.1551455 | 0.0000001 | 0.0094478 | 0.3014689 | NA | NA | 0.3678876 | 0.4365714 | NA |
| Signaling by FGFR in disease.2 | 2 | 4.2e-06 | 0.3600873 | 0.0000003 | 0.0018168 | 0.3913534 | NA | NA | 0.8828437 | 0.8824586 | 0.1272349 |
| Post-translational protein phosphorylation.1 | 1 | 4.3e-06 | 0.0000069 | 0.1293828 | 0.0094200 | 0.6388067 | NA | NA | 0.0646589 | 0.4812814 | 0.0415555 |
| Integrin cell surface interactions.1 | 1 | 4.6e-06 | 0.0000054 | 0.7184853 | 0.0007483 | 0.3736606 | NA | NA | 0.2060560 | 0.3940942 | 0.2297422 |
| Elastic fibre formation.1 | 1 | 4.8e-06 | 0.0003040 | 0.0668308 | 0.0001162 | NA | 0.8977608 | 0.3369076 | 0.1993385 | 0.2497101 | NA |
| NCAM signaling for neurite out-growth.1 | 1 | 4.8e-06 | 0.0000004 | 0.4002290 | 0.0001503 | 0.5139677 | NA | NA | 0.4569624 | 0.0966194 | 0.6087424 |
| Transcriptional Regulation by NPAS4.6 | 6 | 4.8e-06 | 0.0000534 | 0.0843128 | 0.0000060 | NA | NA | NA | NA | 0.0332949 | NA |
| NPAS4 regulates expression of target genes.5 | 5 | 4.8e-06 | 0.0000534 | 0.0843128 | 0.0000060 | NA | NA | NA | NA | 0.0332949 | NA |
| Diseases associated with glycosaminoglycan metabolism.9 | 9 | 4.9e-06 | 0.0000000 | NA | NA | 0.4304541 | NA | NA | NA | 0.8091153 | NA |
| Diseases associated with glycosaminoglycan metabolism.8 | 8 | 5.0e-06 | 0.0000000 | NA | NA | 0.4318175 | NA | NA | NA | 0.9967769 | NA |
Permutations
So far, we have identified some modules that are significant. We can test the robustness of our findings using a resampling procedure. With this strategy, we can repeat the two-class test on a subset of patients, removing 3 random patients at each iteration. Modules are prioritized if they show a resampling success score greater than 80.
We will run the analysis on a subset of pathways, considering only those pathways whose modules were found significant.
Since this step will take much time, results are saved.
useThisPathways <- unique(moduleSummary$pathway[moduleSummary$pvalue <= 0.05])
sModule <- moduleSummary[moduleSummary$pathway %in% useThisPathways, , drop = TRUE]
if (file.exists(paste0(dirname, "permsM.RData"))){
load(paste0(dirname, "permsM.RData"))
}else{
perms <- resamplingModulesTwoClass(fullMultiOmics = multiOmics,
classAnnotation, reactSmall, nperm = 100,
nPatients = 3, pathwaySubset = useThisPathways,
genesToConsider = row.names(pseudoExpNorm))
save(perms, file = paste0(dirname, "permsM.RData"))
}The function selectStablePathwaysModules will retrieve
the tabular summary of the results for those pathway modules whose
success score is greater than 80. It will also print a dotplot showing
the resampling success of modules based on their significance level. The
resampling success count can be retrieved with
getPathwaysModulesSuccess and counts can be appended to the
tabular summary of module results with
addResamplingCounts.
stableModulesSummary <- selectStablePathwaysModules(perms = perms, moduleSummary = sModule, success = 80)
resamplingSuccessCount <- getPathwaysModulesSuccess(perms = perms, moduleSummary = sModule)
moduleSummary <- addResamplingCounts(moduleSummary, resamplingSuccessCount)Plots
The tabular results can also be visualized in a plot, that helps
comparing the contribution of each covariate to the significance of a
module. This is done with plotModuleReport. The user must
provide the MultiOmicsModules object for the pathway of
interest; for this tutorial we choose to focus on the pathway “Syndecan
interactions”.
plotModuleReport(twoClassM[["Syndecan interactions"]],
MOcolors = c(exp = "red", met = "green", cnv = "yellow"),
priority_to = c("exp", "met"))
The most significant module for this pathway is the third one. The omics that are mainly involved are expression and methylation, as we can see from the p-values of the model covariates (expPC1, expPC2, met3k2).
We can have a look at the structure of the pathway graph and the module position in the pathway. The genes of module 3 are colored in red and and the contribution of each omic inside the module is highlighted with different colors.
plotModuleInGraph(twoClassM[["Syndecan interactions"]], reactSmall, 3,
paletteNames = c(exp="red", met="green", cnv="yellow"))
Using a heatmap, we can visualize the profiles of predictive genes, the top 3 genes for each omic. Above the heatmap, we can also show patient additional annotations, in this case their class annotation, and the summarized value of each covariate for each patient.
plotModuleHeat(twoClassM[["Syndecan interactions"]], 3,
additionalAnnotations = classAnnotation,
additionalPaletteNames = list(classes="violet"))
In second instance, we can ask if two or more omics are significant
in the same module simultaneously and if this omic interaction is more
frequent than those expected by chance. To perform this test, we use the
runSupertest function. We will plot only modules that are
significant and have a success score greater than 80. A circle plot is
returned with the frequency of all significant omic combinations and
their significance levels, represented by the height and the color of
the outer layer.
runSupertest(moduleSummary, pvalueThr = 0.05, zscoreThr = 0.05, resampligThr = 80,
excludeColumns = c("pathway", "module", "resamplingCount"))
As you can see, the combination of cnv and expression is significant, as well as the combination of expression and methylation. These combinations of omics co-regulate the same pathway modules more often than what expected by chance.
Finally, with plotFrequencies it is possible to show the
frequency distribution of pathways aggregated into macro-categories,
using Reactome hierarchical structure, separately for each omic
combination. This plot suggests biological processes that may be
impacted by the omics and their cross-talk.
pathHierarchyGraph <- igraph::graph_from_data_frame(d = downloadPathwayRelationFromReactome(), directed = TRUE)
omicsClasses2pathways <- computeOmicsIntersections(moduleSummary,
pvalueThr = 0.05, zscoreThr = 0.05, resampligThr = 80,
excludeColumns = c("pathway", "module", "resamplingCount"))
omicsClasses2pathways <- lapply(omicsClasses2pathways, stripModulesFromPathways)
omicsClasses2fathers <- lapply(omicsClasses2pathways, annotePathwayToFather,
graphiteDB = reactome, hierarchy = pathHierarchyGraph)
MOMfreqDataframe <- computeFreqs(omicsClasses2fathers)
combiClass <- grep(";", MOMfreqDataframe$class)
MOMfreqDataframe.multi <- MOMfreqDataframe[combiClass, , drop = FALSE]
plotFrequencies(MOMfreqDataframe.multi, minSize = 6, maxSize = 8, width = 10, lineSize = 1)
Two-class analysis on pathways
The same analysis can be run on pathways rather than modules. In this
case, since we are using pathway graph structures, we suggest to adopt
the topological method for PCA, as implemented in MOSClip,
as well as a shrinkage approach. To do this, we change parameters for
PCA in the secificArgs slot of our Omics
object.
multiOmics@specificArgs$pcaArgs$shrink=TRUE
multiOmics@specificArgs$pcaArgs$method="topological"We are now ready to run the two-class analysis on pathways with the
function multiOmicsTwoClassPathwayTest. The required input
are the same shown for modules. In this case, we choose to test a
greater amount of pathways (reactHuge list). Again, this
may take some minutes, thus, we save the results
for future usage.
if (file.exists(paste0(dirname, "twoClassP.rds"))){
twoClassP <- readRDS(paste0(dirname, "twoClassP.rds"))
} else {
twoClassP <- lapply(reactHuge, function(g) {
res <- multiOmicsTwoClassPathwayTest(multiOmics, g, classAnnotation,
useTheseGenes = row.names(pseudoExpNorm))
res
})
saveRDS(twoClassP, file = paste0(dirname, "twoClassP.rds"))
}The summary of the results is obtained with
multiPathwayReport.
pathwaySummary <- multiPathwayReport(twoClassP)| cnvNEG | cnvPOS | expPC1 | expPC2 | expPC3 | met2k2 | met3k2 | met3k3 | mut | |
|---|---|---|---|---|---|---|---|---|---|
| Extracellular matrix organization | 0.0240316 | 0.5647068 | 0.0002849 | 0.0188057 | 0.0205115 | 0.0406744 | NA | NA | 0.5778667 |
| Collagen formation | 0.7902942 | 0.7921634 | 0.0000008 | 0.0178320 | 0.8616522 | 0.1180423 | NA | NA | 0.3335339 |
| GPCR downstream signalling | 0.0374317 | 0.3842342 | 0.0002943 | 0.0263907 | 0.0017945 | 0.9299399 | NA | NA | 0.8891513 |
| Signaling by GPCR | 0.0419602 | 0.3749801 | 0.0004191 | 0.0428657 | 0.0012833 | 0.9416765 | NA | NA | 0.9452117 |
| Post-translational protein phosphorylation | 0.1997894 | 0.5651028 | 0.0001413 | 0.0880231 | 0.2291858 | 0.4748054 | NA | NA | 0.0370344 |
| Collagen biosynthesis and modifying enzymes | 0.8417231 | 0.6308409 | 0.0000012 | 0.1424838 | 0.7534697 | 0.7997129 | NA | NA | 0.1864303 |
| Defective B4GALT7 causes EDS, progeroid type | 0.3727263 | 0.9881543 | 0.0000402 | 0.2699674 | 0.1647628 | 0.0462787 | NA | NA | 0.9291883 |
| Defective B3GALT6 causes EDSP2 and SEMDJL1 | 0.3383496 | 0.7703166 | 0.0000366 | 0.2803397 | 0.1423496 | 0.0430582 | NA | NA | 0.9092040 |
| Defective B3GAT3 causes JDSSDHD | 0.3638384 | 0.8173551 | 0.0000343 | 0.2582386 | 0.1577693 | 0.0435254 | NA | NA | 0.9043942 |
| RHO GTPase cycle | 0.0972397 | 0.6429130 | 0.0006720 | 0.0006680 | 0.0005676 | NA | 0.4587927 | 0.1976975 | 0.5002610 |
| Elastic fibre formation | 0.1596668 | 0.0761210 | 0.0001490 | 0.0104862 | 0.5504124 | NA | 0.8148059 | 0.7130069 | 0.2814824 |
| Transcriptional regulation by RUNX2 | 0.0565744 | 0.5169548 | 0.0000111 | 0.0113064 | 0.0013627 | 0.0300799 | NA | NA | 0.2717516 |
| Vesicle-mediated transport | 0.0143504 | 0.6007670 | 0.0000049 | 0.0000001 | 0.1329811 | 0.3858350 | NA | NA | 0.8800403 |
| Integrin cell surface interactions | 0.1817341 | 0.6212187 | 0.0000026 | 0.0175951 | 0.0074610 | NA | 0.0662815 | 0.8565121 | 0.4474758 |
| Regulation of IGF Activity by IGFBP | 0.2302804 | 0.5468850 | 0.0137079 | 0.2038044 | 0.4112719 | NA | 0.3086629 | 0.5264721 | 0.0769803 |
| Chondroitin sulfate/dermatan sulfate metabolism | 0.1695981 | 0.8043619 | 0.0000262 | 0.3117320 | 0.0471199 | 0.0186205 | NA | NA | 0.5189792 |
| A tetrasaccharide linker sequence is required for GAG synthesis | 0.4078676 | 0.9006614 | 0.0000254 | 0.0127004 | 0.6253993 | 0.0093022 | NA | NA | 0.4875072 |
| G alpha (q) signalling events | 0.1554920 | 0.5929441 | 0.0000023 | 0.4096661 | 0.8192870 | 0.3156041 | NA | NA | 0.5945333 |
| Crosslinking of collagen fibrils | 0.2215087 | 0.6214850 | 0.0000256 | 0.0214078 | 0.1356813 | NA | 0.0215082 | 0.5161131 | 0.3401271 |
| Specification of primordial germ cells | NA | 0.9096986 | 0.5943235 | 0.0065019 | 0.0001941 | NA | 0.2214385 | 0.0335146 | NA |
| Axon guidance | 0.0292805 | 0.4144901 | 0.0012508 | 0.0009121 | 0.1040659 | NA | 0.0771731 | 0.1453722 | 0.7220393 |
| ERK1/ERK2 pathway | 0.0434861 | 0.4085110 | 0.0000024 | 0.1184959 | 0.0188482 | 0.8953570 | NA | NA | 0.3097556 |
| Chondroitin sulfate biosynthesis | 0.2028251 | 0.7094322 | 0.0000039 | 0.0033934 | 0.5919170 | 0.0438022 | NA | NA | 0.2142623 |
| Defective EXT2 causes exostoses 2 | 0.4151082 | 0.8819938 | 0.0000001 | 0.4510110 | 0.0004920 | NA | NA | NA | 0.9024605 |
| Defective EXT1 causes exostoses 1, TRPS2 and CHDS | 0.4151082 | 0.8819938 | 0.0000001 | 0.4510110 | 0.0004920 | NA | NA | NA | 0.9024605 |
| PTEN Regulation | 0.0817627 | 0.0829696 | 0.0000149 | 0.0000083 | 0.0002185 | 0.8295500 | NA | NA | 0.3700005 |
| Nervous system development | 0.0314403 | 0.4149122 | 0.0012943 | 0.0010803 | 0.0773709 | NA | 0.0808487 | 0.1760271 | 0.7283126 |
| Diseases associated with glycosaminoglycan metabolism | 0.2914071 | 0.7717944 | 0.8139382 | 0.0001257 | 0.3898244 | NA | 0.7609713 | 0.3264045 | 0.9912342 |
| RAF/MAP kinase cascade | 0.0338995 | 0.3580731 | 0.0000033 | 0.1162157 | 0.0142060 | 0.6604954 | NA | NA | 0.3503695 |
| Carbohydrate metabolism | 0.9130043 | 0.7679701 | 0.0000148 | 0.5077062 | 0.0261936 | 0.7523325 | NA | NA | 0.2194696 |
| RHOB GTPase cycle | 0.0468311 | 0.5121054 | 0.0000001 | 0.0000096 | 0.7671123 | 0.8077366 | NA | NA | 0.8167321 |
| Diseases of glycosylation | 0.2383750 | 0.5069620 | 0.0001157 | 0.0793444 | 0.2916914 | 0.1534655 | NA | NA | 0.6423183 |
| Defective B3GALTL causes PpS | 0.2593327 | 0.8951813 | 0.0000001 | 0.1905794 | 0.1812853 | 0.4831839 | NA | NA | 0.3251967 |
| FGFR2 ligand binding and activation | 0.4558820 | 0.5733507 | 0.0000002 | 0.0000013 | 0.7373796 | 0.0269213 | NA | NA | NA |
| Formation of the ureteric bud | 0.1681610 | 0.1361971 | 0.0000081 | 0.0000116 | 0.0001122 | 0.6744045 | NA | NA | 0.3130224 |
| Signaling by ROBO receptors | 0.1517295 | 0.8897721 | 0.0008698 | 0.0000002 | 0.6001271 | 0.1489777 | NA | NA | 0.2662645 |
| RHOC GTPase cycle | 0.6613835 | 0.6087212 | 0.0000138 | 0.0000130 | 0.0000005 | NA | 0.6975819 | 0.3603495 | 0.0998455 |
| RHOQ GTPase cycle | 0.0962134 | 0.9000713 | 0.0440529 | 0.0000082 | 0.0000003 | 0.7646989 | NA | NA | 0.4050893 |
| Degradation of DVL | 0.1501040 | 0.9367314 | 0.0000039 | 0.0000002 | 0.5887583 | 0.6675524 | NA | NA | 0.8054636 |
| MAPK family signaling cascades | 0.0264043 | 0.5517982 | 0.0000044 | 0.0939433 | 0.0186880 | 0.3147759 | NA | NA | 0.1095113 |
| Diseases associated with O-glycosylation of proteins | 0.4844854 | 0.4118794 | 0.0000001 | 0.7626720 | 0.9848898 | 0.5830885 | NA | NA | 0.4395531 |
| Signaling by FGFR2 in disease | 0.6846049 | 0.7008530 | 0.0000004 | 0.0000007 | 0.5825747 | 0.0208173 | NA | NA | 0.1173073 |
| Phospholipase C-mediated cascade; FGFR2 | 0.7688547 | 0.1901028 | 0.0000004 | 0.0000042 | 0.7770023 | 0.3852312 | NA | NA | NA |
| SHC-mediated cascade:FGFR2 | 0.6254439 | 0.6463056 | 0.0000006 | 0.0000006 | 0.8483130 | 0.0346019 | NA | NA | NA |
| Diseases of signal transduction by growth factor receptors and second messengers | 0.0327482 | 0.3048264 | 0.0000078 | 0.0129148 | 0.0002951 | NA | 0.8535519 | 0.5617953 | 0.5480578 |
| Binding and Uptake of Ligands by Scavenger Receptors | 0.5140621 | 0.7285174 | 0.0001240 | 0.0000109 | 0.0038444 | 0.0292073 | NA | NA | 0.0463858 |
| Regulation of RUNX2 expression and activity | 0.0647823 | 0.4974112 | 0.0000019 | 0.0000680 | 0.0010635 | NA | 0.6550267 | 0.0847859 | 0.0308011 |
| FGFR2c ligand binding and activation | NA | 0.2072975 | 0.0000000 | 0.0012765 | NA | 0.7345982 | NA | NA | NA |
| FGFRL1 modulation of FGFR1 signaling | 0.1028123 | 0.8411326 | 0.0000574 | 0.0000000 | 0.1129987 | 0.4314953 | NA | NA | NA |
| Formation of paraxial mesoderm | 0.0633514 | 0.5488533 | 0.0009316 | 0.0000001 | 0.8402069 | NA | 0.0441185 | 0.9937696 | 0.8711378 |
| Scavenging by Class A Receptors | 0.4051526 | 0.8268658 | 0.0000061 | 0.4822013 | 0.0693799 | 0.7863883 | NA | NA | 0.0162194 |
| Molecules associated with elastic fibres | 0.6220088 | 0.0552455 | 0.0000171 | 0.1844808 | 0.3359382 | NA | 0.9083798 | 0.9522363 | 0.6443183 |
| FGFR2 mutant receptor activation | 0.6594481 | 0.5442979 | 0.0000001 | 0.0000011 | 0.6924788 | 0.1157368 | NA | NA | NA |
| Signaling by PDGF | 0.3355684 | 0.6935265 | 0.0000002 | 0.0354588 | 0.0162812 | 0.2345838 | NA | NA | 0.2626418 |
| Activated point mutants of FGFR2 | 0.7531018 | 0.3331602 | 0.0000004 | 0.0000049 | 0.7920151 | 0.3723379 | NA | NA | NA |
| Dermatan sulfate biosynthesis | 0.2975360 | 0.6157101 | 0.0000068 | 0.1866780 | 0.7230372 | 0.1467363 | NA | NA | 0.5966076 |
| Negative regulation of the PI3K/AKT network | 0.4258167 | 0.0403261 | 0.0000012 | 0.0000498 | 0.7633871 | 0.6932552 | NA | NA | 0.3108647 |
| Trafficking and processing of endosomal TLR | 0.2640648 | 0.8714498 | 0.0000030 | 0.0000134 | 0.0001465 | NA | 0.0014300 | 0.2877000 | 0.5197535 |
| Insulin receptor signalling cascade | 0.2579374 | 0.5197450 | 0.0000013 | 0.0000001 | 0.6301606 | 0.7184197 | NA | NA | 0.6850994 |
| Metabolism of fat-soluble vitamins | 0.0948774 | 0.1201581 | 0.0010215 | 0.0000033 | 0.7597878 | NA | 0.4470069 | 0.0781666 | 0.1167091 |
| Retinoid metabolism and transport | 0.0948774 | 0.1201581 | 0.0010215 | 0.0000033 | 0.7597878 | NA | 0.4470069 | 0.0781666 | 0.1167091 |
| Signaling by FGFR | 0.1288624 | 0.7941618 | 0.0000013 | 0.0000005 | 0.3588054 | NA | 0.6409580 | 0.6371550 | 0.2793489 |
| FRS-mediated FGFR2 signaling | 0.6239538 | 0.5797569 | 0.0000004 | 0.0000004 | 0.9727561 | 0.0660525 | NA | NA | NA |
| Signaling by BMP | 0.4740980 | 0.1027097 | 0.0000012 | 0.0819171 | 0.4171192 | 0.4312054 | NA | NA | 0.5480481 |
| G alpha (i) signalling events | 0.0371717 | 0.4248922 | 0.0000001 | 0.2293518 | 0.0149683 | 0.7057816 | NA | NA | 0.2622769 |
| RHOA GTPase cycle | 0.1024748 | 0.7408258 | 0.0000028 | 0.0000038 | 0.8234946 | NA | 0.6065774 | 0.6057045 | 0.7455201 |
| PI5P, PP2A and IER3 Regulate PI3K/AKT Signaling | 0.3882359 | 0.0312559 | 0.0000020 | 0.0000315 | 0.3688101 | 0.9931072 | NA | NA | 0.2296315 |
| Signaling by Insulin receptor | 0.1166131 | 0.5369190 | 0.0000010 | 0.0000002 | 0.6203859 | 0.8348461 | NA | NA | 0.9889899 |
| Signaling by FGFR2 | 0.2017734 | 0.9313176 | 0.0000007 | 0.0000009 | 0.5888631 | 0.3775783 | NA | NA | 0.2674864 |
| Diseases of metabolism | 0.0839464 | 0.1909300 | 0.0002900 | 0.0676150 | 0.2902320 | NA | 0.0552276 | 0.1123250 | 0.1485337 |
| Negative regulation of FGFR2 signaling | 0.1552913 | 0.7096466 | 0.0000003 | 0.0000007 | 0.9034642 | 0.3689606 | NA | NA | NA |
| Glycosaminoglycan metabolism | 0.7939859 | 0.6189523 | 0.0000023 | 0.1512943 | 0.4365784 | NA | 0.1545235 | 0.3594032 | 0.2511728 |
| PI3K/AKT Signaling in Cancer | 0.2583795 | 0.0314752 | 0.0000021 | 0.0000395 | 0.3146645 | 0.7016615 | NA | NA | 0.5626591 |
| O-glycosylation of TSR domain-containing proteins | 0.3406162 | 0.8548108 | 0.0000004 | 0.1818224 | 0.3893177 | 0.6655576 | NA | NA | 0.3315786 |
| NCAM1 interactions | 0.8840654 | 0.5568446 | 0.0000002 | 0.5580963 | 0.3819905 | 0.4849704 | NA | NA | 0.3705564 |
| The activation of arylsulfatases | 0.5695370 | 0.2168604 | 0.1535106 | 0.0015707 | 0.0000130 | NA | 0.0414831 | 0.4770026 | NA |
| Downstream signal transduction | 0.0177311 | 0.1178433 | 0.0000011 | 0.4141545 | 0.0757079 | 0.6324902 | NA | NA | 0.5933000 |
| PI3K/AKT Signaling | 0.0217046 | 0.1839324 | 0.0000025 | 0.0003501 | 0.5299693 | NA | 0.9728739 | 0.4837122 | 0.3928189 |
| Constitutive Signaling by Aberrant PI3K in Cancer | 0.1164392 | 0.0528870 | 0.0000021 | 0.0000327 | 0.3732082 | 0.9936136 | NA | NA | 0.2683390 |
| Membrane Trafficking | 0.0153472 | 0.3398972 | 0.0000039 | 0.0000001 | 0.0196522 | 0.9839172 | NA | NA | 0.9822560 |
| NCAM signaling for neurite out-growth | 0.4656544 | 0.3854618 | 0.0000004 | 0.1505118 | 0.8890633 | 0.0947353 | NA | NA | 0.8202409 |
| Intra-Golgi and retrograde Golgi-to-ER traffic | 0.1623534 | 0.8255326 | 0.0002684 | 0.1105520 | 0.0000001 | 0.1264353 | NA | NA | 0.4503369 |
| Signaling by TGFB family members | 0.1067845 | 0.9043556 | 0.0000041 | 0.9116613 | 0.9070655 | NA | 0.8827745 | 0.0473354 | 0.1125412 |
| Phospholipase C-mediated cascade; FGFR3 | NA | 0.9029737 | 0.0000000 | 0.0000206 | 0.5856938 | 0.2500034 | NA | NA | NA |
| Synthesis, secretion, and inactivation of Glucagon-like Peptide-1 (GLP-1) | 0.3199954 | 0.9447202 | 0.0001082 | 0.0000143 | 0.0000002 | 0.3221776 | NA | NA | 0.9733780 |
| Regulation of CDH11 Expression and Function | 0.0147802 | 0.4265107 | 0.0000003 | 0.2319873 | 0.7594590 | NA | 0.1508183 | 0.1809470 | 0.4937914 |
| PI-3K cascade:FGFR2 | 0.4293015 | 0.6454894 | 0.0000004 | 0.0000007 | 0.9867624 | 0.4245599 | NA | NA | NA |
| Kidney development | 0.1142341 | 0.8976835 | 0.0000342 | 0.0000038 | 0.0000428 | 0.0600717 | NA | NA | 0.4535685 |
| Integrin signaling | 0.4770905 | 0.6001199 | 0.0000181 | 0.0002779 | 0.0000002 | 0.5051000 | NA | NA | 0.0743639 |
| Visual phototransduction | 0.0656432 | 0.1455824 | 0.1349745 | 0.0000004 | 0.7496279 | 0.0445484 | NA | NA | 0.2201235 |
| RUNX2 regulates osteoblast differentiation | 0.5134977 | 0.8174828 | 0.0000151 | 0.0959687 | 0.0081984 | NA | 0.0357893 | 0.7365349 | 0.5343903 |
| Metabolism of vitamins and cofactors | 0.4158396 | 0.4196352 | 0.0636036 | 0.0000001 | 0.7882397 | 0.7812248 | NA | NA | 0.1809245 |
| FGFR3 ligand binding and activation | NA | 0.8093056 | 0.0000000 | 0.0000381 | 0.7198278 | 0.8511350 | NA | NA | NA |
| FGFR3c ligand binding and activation | NA | 0.8093056 | 0.0000000 | 0.0000381 | 0.7198278 | 0.8511350 | NA | NA | NA |
| Heparan sulfate/heparin (HS-GAG) metabolism | 0.2962408 | 0.8376337 | 0.0000007 | 0.6038985 | 0.0081590 | 0.0164833 | NA | NA | 0.0287701 |
| TCF dependent signaling in response to WNT | 0.0662995 | 0.4546861 | 0.4548073 | 0.0000000 | 0.0440621 | 0.9567258 | NA | NA | 0.5836250 |
| Signaling by Receptor Tyrosine Kinases | 0.0297181 | 0.3435254 | 0.0000003 | 0.0392486 | 0.4509775 | NA | 0.3453504 | 0.9727211 | 0.7256416 |
| Regulation of Homotypic Cell-Cell Adhesion | 0.0359923 | 0.4072230 | 0.2973926 | 0.0000004 | 0.1190335 | NA | 0.2201077 | 0.2469475 | 0.6290851 |
| Regulation of Expression and Function of Type II Classical Cadherins | 0.0359923 | 0.4072230 | 0.2973926 | 0.0000004 | 0.1190335 | NA | 0.2201077 | 0.2469475 | 0.6290851 |
| Syndecan interactions | 0.7773001 | 0.9600632 | 0.0000000 | 0.0011420 | 0.0028720 | NA | 0.7759527 | 0.0289862 | 0.7676395 |
Permutations
Again, we can repeat the test on significant pathways applying a resampling strategy for a total of 100 iteration and we save the results. We can see from the dotplot that the vast majority of pathways is found significant in more than 80 iterations. We can extract stable pathways and add a column with the resampling success count on the pathway summary.
useThisPathways <- unique(row.names(pathwaySummary)[pathwaySummary$pvalue <= 0.05])
sPathway <- pathwaySummary[row.names(pathwaySummary) %in% useThisPathways, , drop = TRUE]
if (file.exists(paste0(dirname, "permsP.RData"))){
load(paste0(dirname, "permsP.RData"))
} else{
perms <- resamplingPathwayTwoClass(fullMultiOmics = multiOmics,
classAnnotation, reactHuge,
nperm = 100, nPatients = 3,
pathwaySubset = useThisPathways,
genesToConsider = row.names(pseudoExpNorm))
save(perms, file = paste0(dirname, "permsP.RData"))
}
stablePathwaysSummary <- selectStablePathwaysModules(perms = perms, moduleSummary = sPathway, success = 80)
resamplingSuccessCount <- getPathwaysModulesSuccess(perms = perms, moduleSummary = sPathway)
pathwaySummary <- addResamplingCounts(pathwaySummary, resamplingSuccessCount)Plots
With the function plotMultiPathwayReport we can plot the
test summary for a subset of pathways.
plotMultiPathwayReport(twoClassP[1:20],
MOcolors = c(exp = "red", mut = "blue",
cnv = "yellow", met = "green"),
priority_to = c("exp", "met"),
fontsize = 5)
We select “Adherens junctions interactions” pathway and plot the heatmap of prioritized genes for each omic.
plotPathwayHeat(twoClassP[["Adherens junctions interactions"]],
additionalAnnotations = classAnnotation,
additionalPaletteNames = list(classes = "violet"))
As found for module tests, the combination of expression and CNV, as well the combination of expression and methylation, co-regulate pathways more often than what expected by chance.
runSupertest(pathwaySummary,
pvalueThr = 0.05, zscoreThr = 0.05, resampligThr = 80,
excludeColumns = "resamplingCount")
The frequency plot shows that the combination of CNV and expression mainly affects signaling pathways, disease and immune system; the same happens with the combination of expression and methylation, with the addition in this case of gene expression macrocategory.
omicsClasses2pathways <- computeOmicsIntersections(pathwaySummary,
pvalueThr = 0.05, zscoreThr = 0.05, resampligThr = 80,
excludeColumns = "resamplingCount")
omicsClasses2fathers <- lapply(omicsClasses2pathways, annotePathwayToFather,
graphiteDB = reactome, hierarchy = pathHierarchyGraph)
freqDataframe <- computeFreqs(omicsClasses2fathers)
combiClass <- grep(";", freqDataframe$class)
freqDataframe.multi <- freqDataframe[combiClass, , drop = FALSE]
plotFrequencies(freqDataframe.multi, minSize = 3, maxSize = 7, width = 9, lineSize = 1)
Data availability
The list of Reactome pathways and the pre-processed datasets used in this tutorial are available here.
Results obtained by running this tutorial are available here.
sessionInfo()## R version 4.4.1 (2024-06-14)
## Platform: x86_64-pc-linux-gnu
## Running under: Ubuntu 22.04.5 LTS
##
## Matrix products: default
## BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.10.0
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
## [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: Europe/Rome
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats4 stats graphics grDevices utils datasets
## [7] methods base
##
## other attached packages:
## [1] kableExtra_1.4.0 MOSClip_0.99.5
## [3] graphite_1.50.0 EDASeq_2.38.0
## [5] ShortRead_1.62.0 GenomicAlignments_1.40.0
## [7] SummarizedExperiment_1.34.0 MatrixGenerics_1.16.0
## [9] matrixStats_1.4.1 Rsamtools_2.20.0
## [11] GenomicRanges_1.56.2 Biostrings_2.72.1
## [13] GenomeInfoDb_1.40.1 XVector_0.44.0
## [15] BiocParallel_1.38.0 org.Hs.eg.db_3.19.1
## [17] AnnotationDbi_1.66.0 IRanges_2.38.1
## [19] S4Vectors_0.42.1 Biobase_2.64.0
## [21] BiocGenerics_0.50.0
##
## loaded via a namespace (and not attached):
## [1] splines_4.4.1 BiocIO_1.14.0
## [3] bitops_1.0-9 ggplotify_0.1.2
## [5] filelock_1.0.3 tibble_3.2.1
## [7] R.oo_1.26.0 qpgraph_2.38.0
## [9] graph_1.82.0 XML_3.99-0.17
## [11] lifecycle_1.0.4 httr2_1.0.5
## [13] rstatix_0.7.2 pwalign_1.0.0
## [15] doParallel_1.0.17 lattice_0.22-6
## [17] MASS_7.3-61 flashClust_1.01-2
## [19] MultiAssayExperiment_1.30.3 SuperExactTest_1.1.0
## [21] NbClust_3.0.1 backports_1.5.0
## [23] magrittr_2.0.3 sass_0.4.9
## [25] rmarkdown_2.28 jquerylib_0.1.4
## [27] yaml_2.3.10 DBI_1.2.3
## [29] RColorBrewer_1.1-3 pkgload_1.4.0
## [31] multcomp_1.4-26 abind_1.4-8
## [33] zlibbioc_1.50.0 purrr_1.0.2
## [35] R.utils_2.12.3 elasticnet_1.3
## [37] RCurl_1.98-1.16 yulab.utils_0.1.7
## [39] TH.data_1.1-2 rappdirs_0.3.3
## [41] sandwich_3.1-1 circlize_0.4.16
## [43] GenomeInfoDbData_1.2.12 KMsurv_0.1-5
## [45] ggrepel_0.9.6 gRbase_2.0.3
## [47] pheatmap_1.0.12 annotate_1.82.0
## [49] svglite_2.1.3 codetools_0.2-20
## [51] DelayedArray_0.30.1 DT_0.33
## [53] xml2_1.3.6 tidyselect_1.2.1
## [55] shape_1.4.6.1 farver_2.1.2
## [57] UCSC.utils_1.0.0 BiocFileCache_2.12.0
## [59] jsonlite_1.8.9 GetoptLong_1.0.5
## [61] Formula_1.2-5 survival_3.7-0
## [63] iterators_1.0.14 emmeans_1.10.5
## [65] coxrobust_1.0.1 systemfonts_1.1.0
## [67] foreach_1.5.2 tools_4.4.1
## [69] progress_1.2.3 Rcpp_1.0.13
## [71] glue_1.8.0 BiocBaseUtils_1.6.0
## [73] gridExtra_2.3 SparseArray_1.4.8
## [75] xfun_0.48 dplyr_1.1.4
## [77] withr_3.0.1 fastmap_1.2.0
## [79] latticeExtra_0.6-30 fansi_1.0.6
## [81] digest_0.6.37 R6_2.5.1
## [83] gridGraphics_0.5-1 estimability_1.5.1
## [85] colorspace_2.1-1 Cairo_1.6-2
## [87] jpeg_0.1-10 biomaRt_2.60.1
## [89] RSQLite_2.3.7 R.methodsS3_1.8.2
## [91] tidyr_1.3.1 utf8_1.2.4
## [93] generics_0.1.3 data.table_1.16.2
## [95] corpcor_1.6.10 rtracklayer_1.64.0
## [97] prettyunits_1.2.0 httr_1.4.7
## [99] htmlwidgets_1.6.4 S4Arrays_1.4.1
## [101] scatterplot3d_0.3-44 pkgconfig_2.0.3
## [103] gtable_0.3.6 blob_1.2.4
## [105] ComplexHeatmap_2.20.0 hwriter_1.3.2.1
## [107] survMisc_0.5.6 htmltools_0.5.8.1
## [109] carData_3.0-5 multcompView_0.1-10
## [111] clue_0.3-65 scales_1.3.0
## [113] leaps_3.2 png_0.1-8
## [115] km.ci_0.5-6 knitr_1.48
## [117] rstudioapi_0.17.1 rjson_0.2.23
## [119] coda_0.19-4.1 checkmate_2.3.2
## [121] curl_5.2.3 cachem_1.1.0
## [123] zoo_1.8-12 GlobalOptions_0.1.2
## [125] stringr_1.5.1 parallel_4.4.1
## [127] restfulr_0.0.15 reshape_0.8.9
## [129] pillar_1.9.0 grid_4.4.1
## [131] vctrs_0.6.5 ggpubr_0.6.0
## [133] car_3.1-3 dbplyr_2.5.0
## [135] xtable_1.8-4 cluster_2.1.6
## [137] Rgraphviz_2.48.0 evaluate_1.0.1
## [139] magick_2.8.5 GenomicFeatures_1.56.0
## [141] mvtnorm_1.3-1 cli_3.6.3
## [143] compiler_4.4.1 rlang_1.1.4
## [145] crayon_1.5.3 ggsignif_0.6.4
## [147] labeling_0.4.3 survminer_0.4.9
## [149] interp_1.1-6 aroma.light_3.34.0
## [151] plyr_1.8.9 fs_1.6.4
## [153] stringi_1.8.4 viridisLite_0.4.2
## [155] deldir_2.0-4 qtl_1.70
## [157] munsell_0.5.1 Matrix_1.7-1
## [159] lars_1.3 hms_1.1.3
## [161] bit64_4.5.2 ggplot2_3.5.1
## [163] KEGGREST_1.44.1 FactoMineR_2.11
## [165] highr_0.11 broom_1.0.7
## [167] igraph_2.1.1 memoise_2.0.1
## [169] bslib_0.8.0 bit_4.5.0