GEOexplorer
1. Introduction
GEOexplorer is a web server and R package for the exploration, generation, and analysis of transcriptomic datasets. GEOexplorer can be used for the exploration, integration, and harmonization of different datasets available in GEO or uploaded by the user. GEOexplorer is based on widely used and validated protocols and enables users to take full advantage of the great availability of high-throughput data both from in-house experiments and publicly available databases. Additionally, GEOexplorer does not require programming proficiency or in-depth statistical knowledge to use.
GEOexplorer enables users to:
Search the GEO database for gene expression datasets.
Retrieve GEO expression datasets or upload their expression datasets.
Merge multiple gene expression datasets and perform batch correction.
Explore gene expression dataset.
Identify the differentially expressed genes in the gene expression dataset.
Perform gene enrichment analysis on the differentially expressed genes.
Explore the results in interactive visualisations.
2. Accessing the GEOexplorer Web Server
GEOexplorer can be accessed on the following link.
3. Alternatively Installing the GEOexplorer R package
GEOexplorer can be installed as an R package from Bioconductor:
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
# The following initializes usage of Bioc devel
BiocManager::install(version='devel')
BiocManager::install("GEOexplorer")Or the latest version can be downloaded from GitHub:
4. Launching the GEOexplorer R package
GEOexplorer can be launched via the two steps below:
Step 1: Load the package
library(GEOexplorer)
#> Loading required package: shiny
#> Loading required package: limma
#> Loading required package: Biobase
#> Loading required package: BiocGenerics
#> Loading required package: generics
#>
#> Attaching package: 'generics'
#> The following objects are masked from 'package:base':
#>
#> as.difftime, as.factor, as.ordered, intersect, is.element, setdiff,
#> setequal, union
#>
#> Attaching package: 'BiocGenerics'
#> The following object is masked from 'package:limma':
#>
#> plotMA
#> The following objects are masked from 'package:stats':
#>
#> IQR, mad, sd, var, xtabs
#> The following objects are masked from 'package:base':
#>
#> Filter, Find, Map, Position, Reduce, anyDuplicated, aperm, append,
#> as.data.frame, basename, cbind, colnames, dirname, do.call,
#> duplicated, eval, evalq, get, grep, grepl, is.unsorted, lapply,
#> mapply, match, mget, order, paste, pmax, pmax.int, pmin, pmin.int,
#> rank, rbind, rownames, sapply, saveRDS, table, tapply, unique,
#> unsplit, which.max, which.min
#> Welcome to Bioconductor
#>
#> Vignettes contain introductory material; view with
#> 'browseVignettes()'. To cite Bioconductor, see
#> 'citation("Biobase")', and for packages 'citation("pkgname")'.
#> Loading required package: plotly
#> Loading required package: ggplot2
#>
#> Attaching package: 'plotly'
#> The following object is masked from 'package:ggplot2':
#>
#> last_plot
#> The following object is masked from 'package:stats':
#>
#> filter
#> The following object is masked from 'package:graphics':
#>
#> layout
#> Loading required package: enrichR
#> Welcome to enrichR
#> Checking connection ...
#> Enrichr ... Connection is Live!
#> FlyEnrichr ... Connection is Live!
#> WormEnrichr ... Connection is Live!
#> YeastEnrichr ... Connection is Live!
#> FishEnrichr ... Connection is Live!
#> OxEnrichr ... Connection is Live!
#> Warning: replacing previous import 'shiny::dataTableOutput' by
#> 'DT::dataTableOutput' when loading 'GEOexplorer'
#> Warning: replacing previous import 'shiny::renderDataTable' by
#> 'DT::renderDataTable' when loading 'GEOexplorer'
#> Setting options('download.file.method.GEOquery'='auto')
#> Setting options('GEOquery.inmemory.gpl'=FALSE)Step 2: Launch the GEOexplorer web application.
3. Tutorial
GEOexplorer splits gene expression analysis into three distinct processes. The first process is exploratory data analysis, which aims to gain an overall understanding of the gene expression dataset. The second process is differential gene expression analysis, which aims to identify the genes that are statistically upregulated or downregulated between two groups. The final process is gene enrichment analysis, which aims to provide the biological context of the differentially expressed genes.
a. GEOexplorer Structure
GEOexplorer contains several tabs, each of which serves a distinct purpose. These tabs will be described in the subsections below.
I. Home Tab
The home tab contains all the widgets to perform gene expression analysis.
II. About Tab
The about tab contains information about GEOexplorer including links to additional documentation.
III. Workflow Tab
The workflow tab provides a high level overview of the workflow performed used by GEOexplorer.
IV. Tutorial Tab
The tutorial tab provides a step by step guide on how to use GEOexplorer.
V. GEO Search Tab
The GEO search tab allows you to search for the GEO database for relevant gene expression datasets to analyse.
VI. Example Datasets Tab
The example datasets tab provides the following: * An example microarray GEO dataset * A gene expression file template * An experimental file template * A microarray gene expression dataset * A microarray experimental conditions file * An RNA-seq gene expression dataset * An RNA-seq experimental conditions file
b. Loading GEO Datasets into GEOexplorer
In this tutorial, we will be exploring the following GEO RNA-seq dataset GSE93939. This dataset contains the gene expression profiles of oculomotor and spinal motor neurons. Oculomotor motor neurons are resilient to degeneration in the lethal motor neuron disease amyotrophic lateral sclerosis (ALS). Therefore comparing the gene expression profiles of oculomotor neurons to spinal motor neurons may indicate the protective mechanisms of oculomotor neurons. There are two ways to automatically load GEO datasets into GEOexplorer. Using the GEO search functionality or manually inputting a GEO accession code
I. Searching for a GEO Dataset
Step 1: Navigate to the GEO Search tab.
Step 2: Input the keywords you which to search. The keywords used in this tutorial are RNA-seq of laser captured oculomotor, cervical and lumbar spinal motor.
Step 3: Select the number of results to display.
Step 4: Click the Search button. A table containing the results will be loaded.
Step 5: Check if the file is processable by GEOexplorer. Due to the variability in the format of GEO RNA-seq datasets, not all GEO RNA-seq datasets can automatically be loaded into GEOexplorer. If the dataset cannot be automatically loaded into GEOexplorer, the user will have to download the dataset and format it into a count matrix as per the template on the Example Datasets tab. However, nearly all GEO microarray datasets can automatically be loaded into GEOexplorer.
Step 6: Click Load Dataset for the microarray gene expression dataset you wish to load. GEOexplorer will attempt to load the dataset from GEO.
II. Using a GEO Accession Code
If you already know the GEO accession code of the dataset you wish to analyse you can perform the following steps.
Step 1: Navigate to the Home tab.
Step 2: Select if you want to analyse multiple datasets or a single dataset. In this example, we will analyse a single dataset.
Step 3: Select the data source. In this example, we will source the dataset directly from GEO.
Step 4: Select if the dataset contains microarray or RNA-seq data.
Step 5: Input the GEO accession code you wish to analyse. GEOexplorer will attempt to load the dataset from GEO. The GEO accession code used in this tutorial is GSE93939.
c. Performing Exploratory Data Analysis
After loading your dataset onto GEOexplorer performing exploratory data analysis is very similar for GEO datasets and user uploaded datasets as well as microarray and RNA-seq datasets.
I. Checking if RNA-seq Datasets Contain Transformed Data
For GEOexplorer to perform differential gene expression analysis RNA-seq datasets must contain the raw counts rather than transformed counts. This step is not required for analysing microarray datasets.
Step 1: If the GEO accession code is linked to multiple datasets please select the platform linked to the dataset you wish to analyse.
Step 2: Select not to apply log transformation. This is important as we want to analyse the non-transformed data.
Step 3: Select not to apply counts per million transformation. This is important as we want to analyse the non-transformed data.
Step 4: Click Analyse.
After clicking “Analyse”, exploratory data analysis will be performed and the results can be reviewed.
Step 5: Click on the Dataset Information tab.
Step 6: Click on the Gene Expression Dataset sub-tab.
Step 7: Review the dataset. If there are any values with non zero decimal places or any negative values, this indicates the RNA-seq dataset has already been transformed and should not be used for differential gene expression analysis.
Step 8: Click on the Exploratory Data Analysis tab.
Step 9: Click on the Expression Density Plot sub-tab.
Step 10: Review the expression density plot. The plot should look similar to the image below. If the plot contains normally distributed density curves with a bell shaped pattern it indicates the RNA-seq dataset has already been transformed and should not be used for differential gene expression analysis.
Step 11: Click on the Box-and-Whisker Plot sub-tab.
Step 12: Review the box-and-whisker plot. The plot should look similar to the image below with the lowest value being 0 or more e.g. a positive number. If the plot contains negative values or median centred values it indicates the RNA-seq dataset has already been transformed and should not be used for differential gene expression analysis.
Note: Whilst transformed datasets should not be used for differential gene expression analysis, they can be used for exploratory data analysis.
II. Reviewing the Results of Exploratory Data Analysis
After checking if the RNA-seq dataset contains transformed data you can continue performing exploratory data analysis.
Step 1: You can set GEOexplorer to perform log transformation or auto-detect if log transformation should be applied.
Step 2: You can set GEOexplorer to perform counts per million transformation. This setting is only available for RNA-seq datasets. For microarray datasets, you can instead select whether to fill in missing values using KNN imputation.
Step 3: Click Analyse to rerun exploratory data analysis.
Step 4: Click on the Dataset Information tab.
Step 5: Review the information presented in the subtabs, which includes details of the experiment, details of the experimental conditions and the gene expression dataset.
Step 6: Click on the Exploratory Data Analysis tab.
Step 7: Review the plots in each of the subtabs. The plots can be divided into four groups:
Group 1: This allows you to identify if the gene expression dataset is normalised. If microarray datasets are not normalised then forced normalisation should be applied during differential gene expression analysis.
Group 2: Displays the amount of variation within each principal component.
Group 3: This allows you to identify if the gene expression dataset contains a large amount of variation. If microarray datasets have a strong mean variance trend then limma precision weights should be applied during differential gene expression analysis.
Group 4: This allows you to identify groups of similar experimental conditions. These different groups can be explored during differential gene expression analysis.
d. Performing Differential Gene Expression Analysis
If the RNA-seq does not contain transformed data differential gene expression analysis. Don’t worry if you applied log transformation or counts per million transformation during exploratory data analysis as GEOexplorer will use the non-log transformed and non-counts per million data.
Step 1: After performing exploratory data analysis, click on the Differential Gene Expression Analysis tab.
Note: As part of differential gene expression analysis you will need to define two groups of experimental conditions you want to compare, to identify the genes that are expressed differently between the two groups.
Step 2: Select the experimental conditions you want to include in Group 1. In the tutorial, we include all the oculomotor neurons in Group 1.
Step 3: Select the experimental conditions you want to include in Group 2. In the tutorial, all the spinal motor neurons in Group 2.
Step 4: Select the P-value adjustment you wish to apply. In the tutorial, the “Benjamini & Hochberg (False discovery rate)” P-value adjustment was selected.
Step 5: Select whether to apply limma precision weights. For RNA-seq datasets, limma precision weights should always be applied. For microarray datasets it is recommended to apply limma precision weights when there is a strong mean-variance trend, as can be identified from the Mean-Variance Plot subtab. Limma precision weights improve the accuracy of differential gene expression analysis when
Step 6: Select whether to force normalisation. For RNA-seq datasets, force normalisation should always be applied. For microarray datasets, force normalisation should be applied if the datasets do not appear normalised from the Box-and-WhisperPlot, Expression Density Plot and 3D Expression Density Plot subtabs.
Step 7: Select the significance level cut off value you want. The cut off will be used to identify the genes that are under-expressed and the genes that are over-expressed between the two groups.
Step 8: Click the Analyse button to perform differential gene expression analysis.
After clicking analyse, differential gene expression analysis will be performed.
Step: 9: Review the results on the following subtabs which can be divided into the following groups:
Group 1: A table containing the statistics of the top 250 differentially expressed genes.
Group 2: Visualisations to indicate if an appropriate P-value adjustment was used.
Group 3: Visualisations that display the differentially expressed genes.
e. Performing Gene Enrichment Analysis
Step 1: After performing differential gene expression analysis, click on the Gene Enrichment Analysis tab.
Step 2: Fill in any missing gene symbols and select the column which contains the gene symbols.
Step 3: Select the database you wish to use to enrich the genes.
Step 4: Click the Analyse button.
After clicking analyse, gene enrichment analysis will be performed.
Step 5: Review the results of gene enrichment analysis.
f. Transforming Datasets into the Format Needed by GEOexplorer
There are two reasons you might need to transform gene expression datasets into a format that GEOexplorer can use. The first reason is that your dataset is not published on GEO. The second reason is that it is published on GEO but cannot be loaded automatically.
I. Identifying that a GEO Dataset Failed to Loaded into GEOexplorer
In this example, we will use the GEO accession code GSE142654.
Step 1: Navigate to the Home tab.
Step 2: If you see the following error “object ‘autoLogInformation’ not found” it indicates something has gone wrong with loading the gene expression dataset.
Step 3: Navigate to the Dataset Information tab.
Step 4: Navigate to the Gene Expression Dataset sub-tab.
Step 5: If the tab is empty it means the gene expression dataset failed to load.
II. Downloading GEO datasets
GEO datasets that fail to automatically load into GEOexplorer need to be downloaded and formatted into the correct format to be processed.
Step 1: Navigate to the GEO Search tab.
Step 2: Enter the GEO accession code or title of the study. In this example, we used GSE142654.
Step 3: Click Search.
Step 4: Click on the GEO link. The GEO website will open in a new tab.
Step 5: Identify which file(s) contain the non-transformed GEO count matrix and click the http link next to the file(s). This will download the file(s).
Step 6: If the file name(s) end in .gz or .tar you will need to ungzip or untar the file(s). There are several websites to do this but these are my favourites: * Ungzip * Untar
Step 7: Back on GEOexplorer, navigate to the Example Datasets tab.
Step 8: Download the *Gene Expression File Template**.
Note: There are several ways you can transform the GEO count matrix file(s) into the GEOexplorer gene expression file template format. However, Excel is by far the simplest. In this example, we will import the GEO count matrix file into Excel.
Step 9: Open the program Excel.
Step 10: Click on the Data tab.
Step 11: Click on the Get Data button.
Step 12: Click on the From File button.
Step 13: Click on the From Text/CSV button.
Step 14: Select the ungzipped GEO count matrix file.
Step 15: Click on the Import button. This step
assumes that Excel correctly identifies the right settings to import the
file. If not, you
need to manually configure these settings.
Step 16: Click on the Load button.
Step 17: Identify the gene ID and gene expression columns from the dataset.
Step 18: Open the GEOexplorer gene expression template in Excel.
Step 19: Copy the gene ID and gene expression columns from the GEO dataset and paste them into the GEOexplorer gene expression template. At this point, I would recommend converting the gene IDs into gene symbols if possible as this will make gene enrichment analysis far easier.
Step 20: Select all the gene expression data and the gene IDs if they are numbers.
Step 21: Click Convert to Number.
Step 22: Save the updated GEOexplorer gene expression template as a CSV.
g. Uploading a Dataset to GEOexplorer
Step 1: Navigate to the Home tab.
Step 2: Select that you want to analyse a single dataset.
Step 3: Select that you want to upload the data yourself.
Step 4: Select if you are uploading RNA-seq or microarray data.
Step 5: Click Browse.
Step 6: Select your gene expression dataset.
Step 7: Click Open.
Step 8: Click Analyse.
Step 9: Continue with your analysis as usual.
6. Reporting Problems or Bugs
If you run into problems using GEOexplorer, the Bioconductor Support site is a good first place to ask for help. If you are convinced that there is a bug in GEOexplorer, feel free to submit an issue on the GEOexplorer GitHub site. Please include the GEO accession code or gene expression dataset that errors, the operating system, and the browser used.
7. Acknowledgements
The development of GEOexplorer was made possible because of the excellent code provided by GEO2R. Additionally, several of GEOexplorer’s key functionalities were enabled because of the R limma package and Enrichr.
8. Session Information
The following package and versions were used in the production of this vignette.
#> R version 4.4.2 (2024-10-31)
#> Platform: x86_64-pc-linux-gnu
#> Running under: Ubuntu 24.04.1 LTS
#>
#> Matrix products: default
#> BLAS: /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
#> LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.26.so; LAPACK version 3.12.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 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: Etc/UTC
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> other attached packages:
#> [1] GEOexplorer_1.13.0 enrichR_3.2 plotly_4.10.4
#> [4] ggplot2_3.5.1 Biobase_2.67.0 BiocGenerics_0.53.3
#> [7] generics_0.1.3 limma_3.63.2 shiny_1.9.1
#> [10] rmarkdown_2.29
#>
#> loaded via a namespace (and not attached):
#> [1] RColorBrewer_1.1-3 sys_3.4.3
#> [3] jsonlite_1.8.9 umap_0.2.10.0
#> [5] magrittr_2.0.3 zlibbioc_1.52.0
#> [7] vctrs_0.6.5 memoise_2.0.1
#> [9] askpass_1.2.1 webshot_0.5.5
#> [11] htmltools_0.5.8.1 S4Arrays_1.7.1
#> [13] curl_6.0.1 cellranger_1.1.0
#> [15] SparseArray_1.7.2 Formula_1.2-5
#> [17] sass_0.4.9 bslib_0.8.0
#> [19] fontawesome_0.5.3 htmlwidgets_1.6.4
#> [21] impute_1.81.0 cachem_1.1.0
#> [23] buildtools_1.0.0 mime_0.12
#> [25] lifecycle_1.0.4 iterators_1.0.14
#> [27] pkgconfig_2.0.3 Matrix_1.7-1
#> [29] R6_2.5.1 fastmap_1.2.0
#> [31] GenomeInfoDbData_1.2.13 MatrixGenerics_1.19.0
#> [33] digest_0.6.37 colorspace_2.1-1
#> [35] AnnotationDbi_1.69.0 S4Vectors_0.45.2
#> [37] shinycssloaders_1.1.0 RSpectra_0.16-2
#> [39] shinybusy_0.3.3 crosstalk_1.2.1
#> [41] RSQLite_2.3.8 GenomicRanges_1.59.1
#> [43] seriation_1.5.6 WriteXLS_6.7.0
#> [45] fansi_1.0.6 mgcv_1.9-1
#> [47] httr_1.4.7 abind_1.4-8
#> [49] compiler_4.4.2 bit64_4.5.2
#> [51] withr_3.0.2 BiocParallel_1.41.0
#> [53] DBI_1.2.3 carData_3.0-5
#> [55] viridis_0.6.5 heatmaply_1.5.0
#> [57] dendextend_1.19.0 R.utils_2.12.3
#> [59] openssl_2.2.2 DelayedArray_0.33.2
#> [61] rjson_0.2.23 tools_4.4.2
#> [63] httpuv_1.6.15 rentrez_1.2.3
#> [65] R.oo_1.27.0 glue_1.8.0
#> [67] nlme_3.1-166 promises_1.3.2
#> [69] grid_4.4.2 sva_3.55.0
#> [71] gtable_0.3.6 tzdb_0.4.0
#> [73] R.methodsS3_1.8.2 ca_0.71.1
#> [75] tidyr_1.3.1 data.table_1.16.2
#> [77] hms_1.1.3 xml2_1.3.6
#> [79] car_3.1-3 utf8_1.2.4
#> [81] XVector_0.47.0 ggrepel_0.9.6
#> [83] foreach_1.5.2 pillar_1.9.0
#> [85] markdown_1.13 stringr_1.5.1
#> [87] genefilter_1.89.0 later_1.4.1
#> [89] splines_4.4.2 dplyr_1.1.4
#> [91] lattice_0.22-6 survival_3.7-0
#> [93] bit_4.5.0 annotate_1.85.0
#> [95] GEOquery_2.75.0 tidyselect_1.2.1
#> [97] registry_0.5-1 locfit_1.5-9.10
#> [99] Biostrings_2.75.1 maketools_1.3.1
#> [101] knitr_1.49 gridExtra_2.3
#> [103] IRanges_2.41.1 edgeR_4.5.0
#> [105] SummarizedExperiment_1.37.0 stats4_4.4.2
#> [107] xfun_0.49 statmod_1.5.0
#> [109] factoextra_1.0.7 matrixStats_1.4.1
#> [111] pheatmap_1.0.12 DT_0.33
#> [113] stringi_1.8.4 UCSC.utils_1.3.0
#> [115] lazyeval_0.2.2 yaml_2.3.10
#> [117] evaluate_1.0.1 codetools_0.2-20
#> [119] tibble_3.2.1 cli_3.6.3
#> [121] xtable_1.8-4 reticulate_1.40.0
#> [123] munsell_0.5.1 jquerylib_0.1.4
#> [125] Rcpp_1.0.13-1 GenomeInfoDb_1.43.2
#> [127] readxl_1.4.3 png_0.1-8
#> [129] parallel_4.4.2 XML_3.99-0.17
#> [131] blob_1.2.4 readr_2.1.5
#> [133] assertthat_0.2.1 shinyHeatmaply_0.2.0
#> [135] viridisLite_0.4.2 scales_1.3.0
#> [137] purrr_1.0.2 crayon_1.5.3
#> [139] rlang_1.1.4 KEGGREST_1.47.0
#> [141] TSP_1.2-4