multiSight quick start guide
26 February 2023
1 multiSight
The purpose of this vignette is to help you become productive as quickly as possible with the multiSight package.
1.1 Version Info
R version: R Under development (unstable) (2023-01-10 r83596)
Bioconductor version:
3.17
Package version:
1.7.0
2 Installation
You can install the released version of multiSight from Bioconductor with:
#To install this package ensure you have BiocManager installed
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
#The following initializes usage of Bioc devel
BiocManager::install("multiSight")3 multiSight purpose
multiSight is an R package providing a graphical interface to analyze and explore your omic datasets (e.g. RNAseq, RPPA) by both single-omics and multi-omics approaches.
Single-omics means Differential Expression Analysis (DEA) is led by DESeq2 towards Over Representation Analysis (ORA). Features with DEA p-values under the 0.05 threshold, or user’s value set in the Biological Insight tab, are selected. These features are saved to carry out ORA on several databases (see Biological insights tab section).
Multi-omics is defined by substituting the DEA part by sPLS-DA method. This is a multi-block selection features method taking several omic blocks into account to select features (see sPLS-DA section). Biological processes (BPs) from a database (e.g. Reactome) are then enriched with these selected features by ORA to reveal putative altered biological processes.
Enrichment results obtained by ORAs are tables with BP IDs, descriptions and p-values. An enrichment table is provided for each omic data block and for each database. multiSight proposes easing enrichment results interpretation from several omic datasets. For each BP shared between enrichment results from a given database, the relative p-values are merged. Stouffer’s pooling method is used to calculate probabilities fusion (see Multi-omics table section). For instance, if “signaling by ERBB2” is highlighted by ORAs of RNAseq and RPPA features with p-values of 0.06 and 0.10, Stouffer’s method calculates a probability of 0.0224 for this pathway. Thereby, multi-omics information is harnessed to make decision of significance.
Stouffer’s p-values for BP beget a multi-omics table where BP IDs, description and Stouffer’s p-values are provided. This multi-omics enrichment table is used to draw an enrichment map.
Enrichment map is a network based on BPs information. BPs are nodes while the bonds are built between BPs sharing some features (see Multi-omics enrichment map section).
In addition, classification models are fitted to select few subsets of features, using biosigner or sPLS-DA methods. biosigner provides one model by omic block and one list of features named biosignature. Nevertheless, sPLS-DA biosignatures are based on more features than biosigner.
Biosignatures can be used:
- To forecast phenotypes (e.g. to diagnostic tasks, histological subtyping);
- To design Pathways and gene ontology enrichments (sPLS-DA biosignatures only);
- To build Network inference;
- To find PubMed references to make assumptions easier and data-driven.
Moreover, numerical relationships between features selected by sPLS-DA or biosigner can feed network inferences (for instance by correlation or mutual information, see Assumption tab section).
multiSight package provides a graphical interface and relative functions to use in a script.
3.1 App structure
multiSight allows to get better biological insights for each omic dataset based on four analytic modules:
- Data input & results;
- Classification models building;
- Biological databases querying;
- Network Inference & PubMed querying.
All figures and tables can be retrieved in an automatic report.
3.2 Which omic data?
All types of omic data respecting input format are supported to build classification models, biosignatures and network inference.
- Genomics;
- Transcriptomics;
- Proteomics;
- Metabolomics;
- Lipidomics;
This package supports all numerical matrices.
You can take a look on data format supported in Home tab details.
4 Home tab
This is the first tab of multiSight.
In this tab, you can load your biological data and find all results in a generated report (.html or .doc formats).
4.1 Data format
You should provide two types of data: numerical matrices and classes vector as csv tables for the same samples.
| Omic data 1 | ||||
|---|---|---|---|---|
| SIGIRR | SIGIRR | MANSC1 | ||
| AOFJ | 0 | 150 | 1004 | … |
| A13E | 34 | 0 | 0 | |
| … |
| Omic data 2 | ||||
|---|---|---|---|---|
| ENSG00000139618 | ENSG00000226023 | ENSG00000198695 | ||
| AOFJ | 25 | 42 | 423 | … |
| A13E | 0 | 154 | 4900 | |
| … |
| Omic classes | |
|---|---|
| Y | |
| AOFJ | condA |
| A13E | condB |
| … |
4.2 Organism
multiSight downloads organism database automatically according to your choice in Home tab. 19 organism databases are available (see Home tab). Please note that multiSight can be used without enrichment analysis, and thus, without a organism database matching with your data.
4.3 Generated report
Report of results can be generated in Home tab below datasets input part inside the Analysis Results block (.html or .doc formats).
5 Classification tab
This tab background is launched when you start analysis in Home tab. Results are displayed after biosigner and sPLS-DA models fitting.
This tab presents classification models to predict sample classes based upon only a small subset of features selected by multiSight models.
5.1 Model methods
Two types of models have been implemented so far to answer different questions: from biosigner & mixOmics (sPLS-DA) R packages.
- To determine small biosignatures - biosigner.
- To build classification models by a multi-block approach - sPLS-DA
- To select relevant biological features to enrich - sPLS-DA
5.1.1 biosigner
biosigner is an R package available in Bioconductor project.
In a single-omics approach, biosigner computes SVM and Random Forest models and selects features for all omic datasets one by one.
Thereby, a biosignature is defined for each omic as the union of SVM and RF variable lists.
If there are 3 omic datasets, biosigner gives 6 models and 3 feature lists (about 5-10 features by data type).
5.1.2 Example
require(multiSight)
## omic2 is a multi-omics dataset of 2 simulated omics included in package
data("omic2", package = "multiSight")
#set.seed(123)
splitData <- splitDatatoTrainTest(omic2, freq = 0.8)
data.train <- splitData$data.train
data.test <- splitData$data.test
## Build model and one biosignature by omic dataset.
biosignerRes <- runSVMRFmodels_Biosigner(data.train)
## Results
biosignerModels <- biosignerRes$model #list of SVM/RF models for each omic.
biosignerFeats <- biosignerRes$biosignature #selected features for each omic.
## Assess model classification performances
biosignerPerf <- assessPerformance_Biosigner(modelList = biosignerModels,
dataTest = data.test)
print(biosignerPerf) #confusion matrices and performance metrics## $svm
## $svm$rnaRead
## Confusion Matrix and Statistics
##
##
## A B
## A 2 0
## B 0 4
##
## Accuracy : 1
## 95% CI : (0.5407, 1)
## No Information Rate : 0.6667
## P-Value [Acc > NIR] : 0.08779
##
## Kappa : 1
##
## Mcnemar's Test P-Value : NA
##
## Sensitivity : 1.0000
## Specificity : 1.0000
## Pos Pred Value : 1.0000
## Neg Pred Value : 1.0000
## Prevalence : 0.3333
## Detection Rate : 0.3333
## Detection Prevalence : 0.3333
## Balanced Accuracy : 1.0000
##
## 'Positive' Class : A
##
##
## $svm$dnaRead
## Confusion Matrix and Statistics
##
##
## A B
## A 2 0
## B 0 4
##
## Accuracy : 1
## 95% CI : (0.5407, 1)
## No Information Rate : 0.6667
## P-Value [Acc > NIR] : 0.08779
##
## Kappa : 1
##
## Mcnemar's Test P-Value : NA
##
## Sensitivity : 1.0000
## Specificity : 1.0000
## Pos Pred Value : 1.0000
## Neg Pred Value : 1.0000
## Prevalence : 0.3333
## Detection Rate : 0.3333
## Detection Prevalence : 0.3333
## Balanced Accuracy : 1.0000
##
## 'Positive' Class : A
##
##
##
## $rf
## $rf$rnaRead
## Confusion Matrix and Statistics
##
##
## A B
## A 2 1
## B 0 3
##
## Accuracy : 0.8333
## 95% CI : (0.3588, 0.9958)
## No Information Rate : 0.6667
## P-Value [Acc > NIR] : 0.3512
##
## Kappa : 0.6667
##
## Mcnemar's Test P-Value : 1.0000
##
## Sensitivity : 1.0000
## Specificity : 0.7500
## Pos Pred Value : 0.6667
## Neg Pred Value : 1.0000
## Prevalence : 0.3333
## Detection Rate : 0.3333
## Detection Prevalence : 0.5000
## Balanced Accuracy : 0.8750
##
## 'Positive' Class : A
##
##
## $rf$dnaRead
## Confusion Matrix and Statistics
##
##
## A B
## A 2 0
## B 0 4
##
## Accuracy : 1
## 95% CI : (0.5407, 1)
## No Information Rate : 0.6667
## P-Value [Acc > NIR] : 0.08779
##
## Kappa : 1
##
## Mcnemar's Test P-Value : NA
##
## Sensitivity : 1.0000
## Specificity : 1.0000
## Pos Pred Value : 1.0000
## Neg Pred Value : 1.0000
## Prevalence : 0.3333
## Detection Rate : 0.3333
## Detection Prevalence : 0.3333
## Balanced Accuracy : 1.0000
##
## 'Positive' Class : A
## 5.1.3 sPLS-DA (DIABLO)
By a multi-omics approach, sPLS-DA (DIABLO) method selects relevant features to explain biological outcome of interest. This is implemented into the mixOmics R package available in Bioconductor project.
sPLS-DA method builds a new space for each omic dataset by linear combinations of initial features giving several components. In fact, around 20-40 features, for each omic, are selected according to contributions to components. Then, these features can be used to enrich some biological processes from a biological database (e.g. KEGG, Reactome) by ORA.
Classification performances are displayed in classification tab for each model and selected features by omic data type (e.g. if you have provided 3 omic types, you could observe 3 feature lists).
5.1.4 Example
require(multiSight)
## omic2 is a multi-omics dataset of 2 simulated omics included in package
data("omic2", package = "multiSight")
data("diabloRes", package = "multiSight")
splitData <- splitDatatoTrainTest(omic2, freq = 0.9)
data.train <- splitData$data.train
data.test <- splitData$data.test## Build model and one biosignature by omic dataset.
# diabloRes <- runSPLSDA(data.train)
# diabloRes #internal object of package to save time
## Results
diabloModels <- diabloRes$model #sPLS-DA model using all omics.
diabloFeats <- diabloRes$biosignature #selected features for each omic.
## Asses model classification performances
#diabloPerf <- assessPerformance_Diablo(splsdaModel = diabloModels,
# dataTest = data.test)
#print(diabloPerf) #confusion matrices and performance metrics6 Biological insights tab
This tab is ready to use when Classification tab features selection is over.
Biological Insights tab is dedicated to give biological sense to your data.
You can carry out enrichment by ORA with features selected according to DESeq2 feature probabilities under the 0.05 threshold. Or, you can compute biological enrichment with features selected by sPLS-DA.
6.1 Biological Annotation Databases
Several databases are implemented in multiSight package to provide a large panel of enrichment analysis:
There are Pathways and Gene Ontology databases helped by clusterProfiler and reactomePA R Bioconductor packages:
- Kegg;
- Reactome;
- wikiPathways;
- Molecular Function (GO)
- Cellular Component (GO)
- Biological Process (GO)
See table below to check available biological information for pathways from these databases, according to organism annotations.
multiSight supports 19 organism databases (required only for enrichment analysis):
| orgDb | kegg | reactome | wikipathways |
|---|---|---|---|
| org.Hs.eg.db | hsa | human | Homo sapiens |
| org.Mm.eg.db | mmu | mouse | Mus musculus |
| org.Rn.eg.db | rno | rat | Rattus norvegicus |
| org.Sc.sgd.db | sce | yeast | Saccharomyces cerevisiae |
| org.Dm.eg.db | dme | fly | Drosophila melanogaster |
| org.Dr.eg.db | dre | zebrafish | Danio rerio |
| org.Ce.eg.db | cel | celegans | Caenorhabditis elegans |
| org.At.tair.db | ath | x | Arabidopsis thaliana |
| org.Bt.eg.db | bta | x | Bos taurus |
| org.Gg.eg.db | gga | x | Gallus gallus |
| org.Cf.eg.db | cfa | x | Canis familiaris |
| org.Ss.eg.db | ssc | x | Sus scrofa |
| org.EcK12.eg.db | eck | x | Escherichia coli |
| org.Pt.eg.db | ptr | x | Pan troglodytes |
| org.Ag.eg.db | aga | x | Anopheles gambiae |
| org.Pf.plasmo.db | pfa | x | Plasmodium falciparum |
| org.EcSakai.eg.db | ecs | x | Escherichia coli |
| org.Mmu.eg.db | mcc | x | x |
| org.Xl.eg.db | xla | x | x |
Note for enrichment: only convertible feature names to genes could be enriched according to information in organism’s database (e.g. Gene SYMBOL to entrez ids).
To know what feature name types are supported, copy these two lines of codes with your organism database of interest. For instance, org.Mm.eg.db for Human provides 24 input names you can use as feature names in your datasets.
if (requireNamespace("org.Mm.eg.db", quietly = TRUE))
{
library(org.Mm.eg.db)
columns(org.Mm.eg.db)
}## [1] "ACCNUM" "ALIAS" "ENSEMBL" "ENSEMBLPROT" "ENSEMBLTRANS"
## [6] "ENTREZID" "ENZYME" "EVIDENCE" "EVIDENCEALL" "GENENAME"
## [11] "GENETYPE" "GO" "GOALL" "IPI" "MGI"
## [16] "ONTOLOGY" "ONTOLOGYALL" "PATH" "PFAM" "PMID"
## [21] "PROSITE" "REFSEQ" "SYMBOL" "UNIPROT"Please note that feature IDs can differ between omic blocks (e.g. one dataset with gene SYMBOL, another one with ENSEMBL IDs). You can provide this information to convert them for enrichment in the application.
6.2 Single-omic DESeq2 differential expression analysis
Therefore, in this Biological insights tab you can compute Differential Expression Analysis (DEA) by DESeq2 for several omic datasets Moreover, with features selected according to p-value threshold set by the user you can enrich pathway or Gene Ontology databases.
6.2.1 Example
require(multiSight)
## omic2 is a multi-omics dataset of 2 simulated omics included in package
data("omic2", package = "multiSight")
data("deseqRes", package = "multiSight")
# deseqRes <- runMultiDeseqAnalysis(omicDataList = omic2,
# padjUser = 0.05)
## One Differential Expression Analysis table for each omic dataset
# View(deseqRes$DEtable)
## One feature selected list for each omic according to padjust user threshold
multiOmic_biosignature <- deseqRes$selectedFeatures
# View(multiOmic_biosignature)
## Multi-omics enrichment
### convert features
dbList <- list(rnaRead = "ENSEMBL",
dnaRead = "ENSEMBL")
convFeat <- convertToEntrezid(multiOmic_biosignature, dbList, "org.Mm.eg.db")
### ORA enrichment analysis
if (requireNamespace("org.Mm.eg.db", quietly = TRUE))
{
# database <- c("kegg", "wikiPathways", "reactome", "MF", "CC", "BP")
database <- c("reactome")
data("enrichResList", package = "multiSight")
# enrichResList <- runMultiEnrichment(omicSignature = convFeat,
# databasesChosen = database,
# organismDb = "org.Mm.eg.db",
# pvAdjust = "BH", #default value
# minGSSize = 5, #default value
# maxGSSize = 800, #default value
# pvStouffer = 0.1) #default value
reacRes <- enrichResList$pathways$reactome
names(reacRes$result) # classical enrichment tables, multi-omics and EMap
}## NULL6.3 Multi-omics sPLS-DA selected features
This tab gives a graphical interface where all multi-omics features selected by sPLS-DA method are used to enrich chosen databases. You could retrieve functional enrichment results in Home tab’s report or save tables in this tab.
require(multiSight)
## omic2 is a multi-omics dataset of 2 simulated omics included in package
data("omic2", package = "multiSight")
data("diabloRes", package = "multiSight")
# splitData <- splitDatatoTrainTest(omic2, 0.8)
# data.train <- splitData$data.train
# data.test <- splitData$data.test
#
# diabloRes <- runSPLSDA(data.train)
# diabloRes #internal object of package to save time
diabloModels <- diabloRes$model #sPLS-DA model using all omics.
diabloFeats <- diabloRes$biosignature #selected features for each omic.
## Multi-omics enrichment
### convert features
names(diabloFeats) #/!\use same names for dbList and omic datasets.## [1] "rnaRead" "dnaRead"dbList <- list(rnaRead = "ENSEMBL", #feature names origin
dnaRead = "ENSEMBL")
if (requireNamespace("org.Mm.eg.db", quietly = TRUE))
{
convFeat <- convertToEntrezid(diabloFeats,
dbList,
"org.Mm.eg.db")
### ORA enrichment analysis for omic feature lists
# database <- c("kegg", "wikiPathways", "reactome", "MF", "CC", "BP")
database <- c("reactome")
multiOmicRes <- runMultiEnrichment(omicSignature = convFeat,
databasesChosen = database,
organismDb = "org.Mm.eg.db",
pvAdjust = "BH", #default value
minGSSize = 5, #default value
maxGSSize = 800, #default value
pvStouffer = 0.1) #default value
## Results
reacRes <- multiOmicRes$pathways$reactome
names(reacRes$result) # classical enrichment tables, multi-omics and EMap
}## [1] "enrichTable_omic1" "enrichTable_omic2" "multi"
## [4] "enrichMap"6.4 Visualizations
Two types of result visualization are given:
- Classical enrichment tables for each omic and each database (with pathways id, p-value, padjust column).
- Then, when more than one omic is used for enrichment, a multi-omics table and a multi-omics enrichment map with DEA and sPLS-DA selected features are available.
6.4.1 Multi-omics Table
A multi-omics table is built by annotation database (e.g. Reactome) with all enrichment results obtained for omic datasets using Stouffer’s method. Stouffer’s p-value method consists in p-values pooling for same pathways or ontology. In fact, a Stouffer’s value is computed for every biological annotations shared by at least 2 enrichment analysis results (NOTE: each omic data used for enrichment has one usual enrichment table by selected database).
Thereby, you can summarize information for several datasets and enrichment results with a multi-omics table, for instance:
| ID | Description | p-value:Omic1 | p-value:Omic2 | Stouffer | StoufferWeighted | geneID | GeneRatio | Count |
|---|---|---|---|---|---|---|---|---|
| R-MMU-2173791 | TGF-beta receptor signaling in EMT (epithelial to mesenchymal transition) | 0.043 | 0.047 | 0.008 | 0.008 | 102098/11848/16456/18762/21803/… | 2/60 | 2 |
| R-MMU-194315 | Signaling by Rho GTPases | 0.096 | 0.029 | 0.011 | 0.011 | 100043813/101497/102098/102920/… | 1/60 | 1 |
| R-MMU-73887 | Death Receptor Signalling | 0.056 | 0.06 | 0.013 | 0.013 | 101497/102098/106025/109934/11491/ | 2/60 | 2 |
| R-MMU-8953854 | Metabolism of RNA | 0.533 | 0.002 | 0.024 | 0.021 | 100043813/100044627/100502825/… | 1/60 | 1 |
Each value is similar to an enrichRes object obtained by clusterProfiler.
In fact, Stouffer’s results are transformed as *enrichRes** objects and could be used for usual enrichRes analysis and visualization functions.
Note in our cytoscape-like case that geneID column refers to all genes in pathways or ontology named in database annotations. In usual enrichRes, this column presents genes enriched contributing in pathways.
A graphical network is built according to this table as an Enrichment Map.
6.4.2 Multi-omics enrichment map
Enrichment map is a graph drawn to observe pathway or ontology relationships according to overlapping elements in each parts.
Each pathway is a node and a bond is built between them if there have common genes (see Jaccard similarity coefficient - JC’s similarity method).
Here, enrichment map functions from enrichPlot package are adapted to draw similar map to cytoscape add-on EnrichmentMap: all genes in each pathway are used to compute similarity between pathway building network bonds.
Each multi-omics enrichment map
(one by enrichment database) could be retrieve in a report
automatically generated and in the Home tab to be saved.
7 Assumption tab
Assumption tab aims to help biological hypothesis making by network inference from feature relationship values (e.g correlation, partial correlation) and by a PubMed module.
You can find both functions:
- To compute network inference and to reveal feature relationships.
- To get PubMed articles based on your personalized query without leaving app.
require(multiSight)
data("omic2", package = "multiSight")
data("diabloRes", package = "multiSight")
## omic2 is a multi-omics dataset of 2 simulated omics included in package
splitData <- splitDatatoTrainTest(omic2, 0.8)
data.train <- splitData$data.train
data.test <- splitData$data.test
## Build sPLS-DA models
# diabloRes <- runSPLSDA(data.train)
diabloFeats <- diabloRes$biosignature #selected features for each omic.
## Build biosigner models
biosignerRes <- runSVMRFmodels_Biosigner(data.train)
biosignerFeats <- biosignerRes$biosignature #selected features for each omic.
## Network inference
### DIABLO features
concatMat_diablo <- getDataSelectedFeatures(omic2, diabloFeats)
corrRes_diablo <- correlationNetworkInference(concatMat_diablo, 0.85)
pcorRes_diablo <- partialCorrelationNI(concatMat_diablo, 0.4)
miRes_diablo <- mutualInformationNI(concatMat_diablo, 0.2)
### biosigner features
concatMat_biosigner <- getDataSelectedFeatures(omic2, biosignerFeats)
corrRes_bios <- correlationNetworkInference(concatMat_biosigner, 0.85)
pcorRes_bios <- partialCorrelationNI(concatMat_biosigner, 0.4)
miRes_bios <- mutualInformationNI(concatMat_biosigner, 0.2)
corrRes_diablo$graphSession info
Here is the output of sessionInfo() on the system on which this
document was compiled running pandoc 2.9.2.1:
## R Under development (unstable) (2023-01-10 r83596)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 22.04.1 LTS
##
## Matrix products: default
## BLAS: /home/biocbuild/bbs-3.17-bioc/R/lib/libRblas.so
## 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_GB LC_COLLATE=C
## [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: America/New_York
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats4 stats graphics grDevices utils datasets methods
## [8] base
##
## other attached packages:
## [1] org.Mm.eg.db_3.16.0 AnnotationDbi_1.61.0 IRanges_2.33.0
## [4] S4Vectors_0.37.4 Biobase_2.59.0 BiocGenerics_0.45.0
## [7] multiSight_1.7.0 BiocStyle_2.27.1
##
## loaded via a namespace (and not attached):
## [1] fs_1.6.1 matrixStats_0.63.0
## [3] bitops_1.0-7 enrichplot_1.19.1
## [5] lubridate_1.9.2 HDO.db_0.99.1
## [7] httr_1.4.5 RColorBrewer_1.1-3
## [9] numDeriv_2016.8-1.1 tools_4.3.0
## [11] utf8_1.2.3 R6_2.5.1
## [13] DT_0.27 sn_2.1.0
## [15] lazyeval_0.2.2 withr_2.5.0
## [17] graphite_1.45.0 gridExtra_2.3
## [19] cli_3.6.0 scatterpie_0.1.8
## [21] sandwich_3.0-2 labeling_0.4.2
## [23] sass_0.4.5 mvtnorm_1.1-3
## [25] randomForest_4.7-1.1 proxy_0.4-27
## [27] yulab.utils_0.0.6 gson_0.0.9
## [29] DOSE_3.25.0 parallelly_1.34.0
## [31] plotrix_3.8-2 attempt_0.3.1
## [33] limma_3.55.4 rstudioapi_0.14
## [35] RSQLite_2.3.0 generics_0.1.3
## [37] gridGraphics_0.5-1 dplyr_1.1.0
## [39] GO.db_3.16.0 Matrix_1.5-3
## [41] qqman_0.1.8 fansi_1.0.4
## [43] infotheo_1.2.0.1 lifecycle_1.0.3
## [45] multcomp_1.4-22 yaml_2.3.7
## [47] mathjaxr_1.6-0 SummarizedExperiment_1.29.1
## [49] recipes_1.0.5 qvalue_2.31.1
## [51] grid_4.3.0 blob_1.2.3
## [53] promises_1.2.0.1 ppcor_1.1
## [55] crayon_1.5.2 shinydashboard_0.7.2
## [57] lattice_0.20-45 MultiDataSet_1.27.0
## [59] cowplot_1.1.1 KEGGREST_1.39.0
## [61] pillar_1.8.1 knitr_1.42
## [63] fgsea_1.25.0 GenomicRanges_1.51.4
## [65] rjson_0.2.21 corpcor_1.6.10
## [67] mixOmics_6.23.4 future.apply_1.10.0
## [69] codetools_0.2-19 fastmatch_1.1-3
## [71] mutoss_0.1-12 glue_1.6.2
## [73] downloader_0.4 ggfun_0.0.9
## [75] data.table_1.14.8 MultiAssayExperiment_1.25.7
## [77] Rdpack_2.4 vctrs_0.5.2
## [79] png_0.1-8 treeio_1.23.0
## [81] gtable_0.3.1 cachem_1.0.7
## [83] gower_1.0.1 xfun_0.37
## [85] rbibutils_2.2.13 mime_0.12
## [87] prodlim_2019.11.13 tidygraph_1.2.3
## [89] survival_3.5-3 rWikiPathways_1.19.0
## [91] timeDate_4022.108 iterators_1.0.14
## [93] hardhat_1.2.0 lava_1.7.2
## [95] TH.data_1.1-1 ellipsis_0.3.2
## [97] ipred_0.9-13 nlme_3.1-162
## [99] ggtree_3.7.1 usethis_2.1.6
## [101] bit64_4.0.5 GenomeInfoDb_1.35.15
## [103] rprojroot_2.0.3 bslib_0.4.2
## [105] rpart_4.1.19 colorspace_2.1-0
## [107] DBI_1.1.3 nnet_7.3-18
## [109] mnormt_2.1.1 DESeq2_1.39.6
## [111] tidyselect_1.2.0 bit_4.0.5
## [113] compiler_4.3.0 graph_1.77.2
## [115] TFisher_0.2.0 xml2_1.3.3
## [117] desc_1.4.2 DelayedArray_0.25.0
## [119] bookdown_0.32 shadowtext_0.1.2
## [121] scales_1.2.1 rappdirs_0.3.3
## [123] stringr_1.5.0 digest_0.6.31
## [125] rmarkdown_2.20 XVector_0.39.0
## [127] htmltools_0.5.4 pkgconfig_2.0.3
## [129] MatrixGenerics_1.11.0 fastmap_1.1.1
## [131] rlang_1.0.6 htmlwidgets_1.6.1
## [133] shiny_1.7.4 farver_2.1.1
## [135] jquerylib_0.1.4 zoo_1.8-11
## [137] jsonlite_1.8.4 BiocParallel_1.33.9
## [139] GOSemSim_2.25.0 ropls_1.31.0
## [141] ModelMetrics_1.2.2.2 config_0.3.1
## [143] RCurl_1.98-1.10 magrittr_2.0.3
## [145] GenomeInfoDbData_1.2.9 ggplotify_0.1.0
## [147] patchwork_1.1.2 munsell_0.5.0
## [149] Rcpp_1.0.10 ggnewscale_0.4.8
## [151] ape_5.7 viridis_0.6.2
## [153] anyLib_1.0.5 stringi_1.7.12
## [155] pROC_1.18.0 ggraph_2.1.0
## [157] zlibbioc_1.45.0 MASS_7.3-58.2
## [159] plyr_1.8.8 parallel_4.3.0
## [161] listenv_0.9.0 ggrepel_0.9.3
## [163] Biostrings_2.67.0 graphlayouts_0.8.4
## [165] splines_4.3.0 multtest_2.55.0
## [167] qqconf_1.3.1 locfit_1.5-9.7
## [169] igraph_1.4.1 reshape2_1.4.4
## [171] pkgload_1.3.2 XML_3.99-0.13
## [173] evaluate_0.20 metap_1.8
## [175] calibrate_1.7.7 golem_0.3.5
## [177] BiocManager_1.30.20 foreach_1.5.2
## [179] tweenr_2.0.2 httpuv_1.6.9
## [181] networkD3_0.4 tidyr_1.3.0
## [183] purrr_1.0.1 polyclip_1.10-4
## [185] future_1.31.0 ggplot2_3.4.1
## [187] ReactomePA_1.43.0 ggforce_0.4.1
## [189] xtable_1.8-4 reactome.db_1.82.0
## [191] easyPubMed_2.13 RSpectra_0.16-1
## [193] e1071_1.7-13 tidytree_0.4.2
## [195] roxygen2_7.2.3 later_1.3.0
## [197] viridisLite_0.4.1 class_7.3-21
## [199] rARPACK_0.11-0 tibble_3.1.8
## [201] clusterProfiler_4.7.0 aplot_0.1.9
## [203] ellipse_0.4.3 memoise_2.0.1
## [205] biosigner_1.27.0 timechange_0.2.0
## [207] globals_0.16.2 caret_6.0-93