TissueEnrich

The TissueEnrich package is used to calculate enrichment of tissue-specific genes in a set of input genes. For example, the user can input the most highly expressed genes from RNA-Seq data, or gene co-expression modules to determine which tissue-specific genes are enriched in those datasets. Tissue-specific genes were defined by processing RNA-Seq data from the Human Protein Atlas (HPA) (Uhlén et al. 2015), GTEx (Ardlie et al. 2015), and mouse ENCODE (Shen et al. 2012) using the algorithm from the HPA (Uhlén et al. 2015). The hypergeometric test is being used to determine if the tissue-specific genes are enriched among the input genes. Along with tissue-specific gene enrichment, the TissueEnrich package can also be used to define tissue-specific genes from expression datasets provided by the user, which can then be used to calculate tissue-specific gene enrichments. TissueEnrich has the following three functions.

How to get help for TissueEnrich

Please post all the questions or queries related to TissueEnrich package on the Bioconductor support website. This will help us to build an information repository which can be used by other users.

https://support.bioconductor.org

Please do not email your questions directly to the package authors.

teEnrichment: Tissue-specific gene enrichment using human or mouse genes

The teEnrichment function is used to calculate the enrichment of tissue-specific genes in an input gene set. It uses tissue-specific genes defined by processing RNA-Seq datasets from human and mouse. The user must specify the organism using the organism (“Homo Sapiens” (default) or “Mus Musculus”) parameter in the input GeneSet object. More details about the RNA-Seq datasets and tissue-specific genes are discussed in the next sections.

RNA-Seq datasets

TissueEnrich defines tissue-specific genes using RNA-Seq data from the HPA, GTEx, and mouse ENCODE. In order to make the tissue-specific gene calculations more robust, we only used tissues that had ≥2 biological replicates. The datasets used in the tool are:

  • HPA Dataset: RNA-Seq data across 35 human tissues (Uhlén et al. 2015).
  • GTEx Dataset: RNA-Seq data across 29 human tissues (Ardlie et al. 2015).
  • Mouse ENCODE Dataset: RNA-Seq data across 17 mouse tissues (Shen et al. 2012).

When using teEnrichment, the user can specify the RNA-Seq dataset (rnaSeqDataset) to be used for the tissue-specific gene enrichment analysis.

  • 1 for “Human Protein Atlas” (default)
  • 2 for “GTEx”
  • 3 for “Mouse ENCODE”

Note: The tissues isolated from embryonic stages are prefixed with ‘E’ followed by the timepoint. For example in Mouse ENCODE data, the placenta tissue isolated from embryonic day 14.5 is named as E14.5-Placenta. All the other tissues are isolated from adults.

Defining Tissue-specific Genes

Tissue-specific genes are defined using the algorithm from the HPA (Uhlén et al. 2015), and can be grouped as follows:

  • Tissue Enriched: Genes with an expression level greater than 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a particular tissue compared to all other tissues.
  • Group Enriched: Genes with an expression level greater than 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a group of 2-7 tissues compared to all other tissues, and that are not considered Tissue Enriched.
  • Tissue Enhanced: Genes with an expression level greater than 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a particular tissue compared to the average levels in all other tissues, and that are not considered Tissue Enriched or Group Enriched.

In teEnrichment, the user can specify the type of tissue-specific genes (tissueSpecificGeneType) to be used for the tissue-specific gene enrichment analysis.

  • 1 for “All” (default)
  • 2 for “Tissue-Enriched”
  • 3 for “Tissue-Enhanced”
  • 4 for “Group-Enriched”

Hypergeometric test

The hypergeometric test is used to calculate tissue-specific gene enrichment. The p-value is calculated as:

\[ P(X \gt k) = \sum\limits_{i=k+1}^n \frac{{{K}\choose{i}} {{N-K}\choose{n-i}}}{{{N}\choose{n}}} \] and the fold-change is calculated as: \[ Fold-change =\left( \frac{k}{n} \right)/\left( \frac{K}{N} \right) \]

Where, N is the total number of genes, K is the total number of tissue-specific genes for a tissue, n is the number of genes in the input gene set, k is the number of tissue-specific genes in the input gene set. The p-values can be corrected for multiple hypothesis testing using the Benjamini & Hochberg correction by setting multiHypoCorrection = TRUE (It is TRUE by default).

Background genes

TissueEnrich now enables users to provide the background genes for carrying out tissue-specific gene enrichment. In this case, instead of using all the genes in the dataset, a background gene set is being used to carry out the enrichment analysis. It should be noted that the background gene set must have all the genes of the input gene set. The p-value is calculated as: \[ P(X \gt k) = \sum\limits_{i=k+1}^n \frac{{{K_b}\choose{i}} {{N_b-K_b}\choose{n-i}}}{{{N_b}\choose{n}}} \] and the fold-change is calculated as: \[ Fold-change =\left( \frac{k}{n} \right)/\left( \frac{K_b}{N_b} \right) \] Where, \(N_b\) is the total number of background genes, \(K_b\) is the total number of tissue-specific genes for a tissue in background genes, n is the number of genes in the input gene set, k is the number of tissue-specific genes in the input gene set. The p-values can be corrected for multiple hypothesis testing using the Benjamini & Hochberg correction by setting multiHypoCorrection = TRUE (It is TRUE by default). If the background gene set is not provided all the genes will be used as background.

Example: Tissue-specific gene enrichment

This example uses trophectoderm (TE) specific genes identified from single cell RNA-Seq analyses, performed on human blastocysts on days 5, 6, and 7 of preimplantation development (Petropoulos et al. 2016). The single cells are assigned to either the inner cell mass (epiblast plus emerging extraembryonic endoderm) or the TE using PCA. After that, a list of 100 TE-specific genes was generated using differential gene expression analysis (Petropoulos et al. 2016). We used those 100 genes as the input gene set and carried out tissue-specific gene enrichment using the tissue-specific genes defined by the HPA dataset.

Note: The input gene set can either contain Ensembl Ids (ENSEMBLIdentifier()) or Gene Symbols (SymbolIdentifier()) (specify this using the geneIdType parameter in the input GeneSet object).

The output is a list object containing the enrichment results. These results are explained in the next section.

Exploring tissue-specific gene enrichment results

Tissue-specific gene enrichment bar chart using ggplot2

The first object in the output list is a SummarizedExperiment object containing the \(-Log_{10}(P-Value)\) and fold-change, corresponding to the tissue-specific gene enrichment, along with the number of tissue-specific genes in the input gene set. This object can be used to visualize tissue-specific gene enrichment in the form of a bar chart using the \(-Log_{10}(P-Value)\) values.

In the plot above, the x-axis shows each of the tissues, and the y-axis represents the tissue-specific gene enrichment (\(-Log_{10}(P-Value)\)) values. As expected, the 100 TE-specific genes show enrichment for placenta specific genes.

This output object used to visualize tissue-specific gene enrichment using the \(-Log_{10}(P-Value)\) values can also be used to plot the fold-change values.

In the plot above, the x-axis shows each of the tissues, and the y-axis represents the fold-change values of the tissue-specific gene enrichment.

Retrieval of input tissue-specific genes

The third object in the output is a list containing the tissue-specificity information for the input genes. The code below retrieves the tissue-specific genes along with the type of tissue-specificity in placenta tissue.

Retrieval of tissue-specific genes that could not be mapped

The fourth object in the output list is a character vector that has a list of input genes that were not identified in the tissue-specific gene data.

Orthologous gene enrichment

The teEnrichment function can calculate mouse tissue-specific gene enrichment from a human gene list or vice versa. The user simply specifies if the input data is from mouse or human, and selects the tissue-specific gene data of interest, whether it is from mouse or human. The function will automatically carry out orthologous tissue-specific gene enrichment using one-to-one protein coding orthologous genes between human and mouse, downloaded from Ensembl V91 database (Aken et al. 2016).

teGeneRetrieval: Identification of tissue-specific genes

The teGeneRetrieval function is used to define tissue-specific genes, using the algorithm from the HPA (Uhlén et al. 2015). It takes an SummarizedExperiment object containing expression information as input (rows as genes and columns as tissue) and classifies the genes into different gene groups and returns the information in another SummarizedExperiment object. The users also have the options of changing the default thresholds to vary the degree of tissue specificity of genes. More details about the gene groups and HPA thresholds are provided below.

Gene groups

The genes are divided into six groups based on their gene expression across the tissues. These groups are:

  • Not Expressed: Genes with an expression level less than 1 (TPM or FPKM) across all the tissues.
  • Tissue Enriched: Genes with an expression level greater than or equal to 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a particular tissue compared to all other tissues.
  • Group Enriched: Genes with an expression level greater than or equal to 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a group of 2-7 tissues compared to all other tissues, and that are not considered Tissue Enriched.
  • Tissue Enhanced: Genes with an expression level greater than or equal to 1 (TPM or FPKM) that also have at least five-fold higher expression levels in a particular tissue compared to the average levels in all other tissues, and that are not considered Tissue Enriched or Group Enriched.
  • Expressed in all: Genes with an expression level greater than or equal to 1 (TPM or FPKM) across all of the tissues that are not in any of the above 4 groups.
  • Mixed: Genes that are not assigned to any of the above 5 groups.

Genes from the Tissue Enriched, Group Enriched, and Tissue Enhanced groups are classified as tissue-specific genes.

teEnrichmentCustom: Tissue-specific gene enrichment in custom expression datasets

The teEnrichmentCustom function is used to calculate tissue-specific gene enrichment using tissue-specific genes defined using the teGeneRetrieval function.

References

Aken, Bronwen L, Sarah Ayling, Daniel Barrell, Laura Clarke, Valery Curwen, Susan Fairley, Julio Fernandez Banet, et al. 2016. “The Ensembl gene annotation system.” Database : The Journal of Biological Databases and Curation 2016. Oxford University Press. https://doi.org/10.1093/database/baw093.

Ardlie, Kristin G., David S. Deluca, Ayellet V. Segrè, Timothy J. Sullivan, Taylor R. Young, Ellen T. Gelfand, Casandra A. Trowbridge, et al. 2015. “The Genotype-Tissue Expression (Gtex) Pilot Analysis: Multitissue Gene Regulation in Humans.” Science 348 (6235). American Association for the Advancement of Science:648–60. https://doi.org/10.1126/science.1262110.

Petropoulos, Sophie, Daniel Edsgärd, Björn Reinius, Qiaolin Deng, Sarita Pauliina Panula, Simone Codeluppi, Alvaro Plaza Reyes, Sten Linnarsson, Rickard Sandberg, and Fredrik Lanner. 2016. “Single-Cell RNA-Seq Reveals Lineage and X Chromosome Dynamics in Human Preimplantation Embryos.” Cell 165 (4). Elsevier:1012–26. https://doi.org/10.1016/j.cell.2016.03.023.

Shen, Yin, Feng Yue, David F. McCleary, Zhen Ye, Lee Edsall, Samantha Kuan, Ulrich Wagner, et al. 2012. “A map of the cis-regulatory sequences in the mouse genome.” Nature 448 (7409). http://www.nature.com/articles/nature11243.

Uhlén, Mathias, Linn Fagerberg, Björn M. Hallström, Cecilia Lindskog, Per Oksvold, Adil Mardinoglu, Asa Sivertsson, et al. 2015. “Tissue-Based Map of the Human Proteome.” Science 347 (6220). American Association for the Advancement of Science. https://doi.org/10.1126/science.1260419.