In lgatto/hpar: Human Protein Atlas in R

suppressPackageStartupMessages(library("dplyr"))
suppressPackageStartupMessages(library("BiocStyle"))
suppressPackageStartupMessages(library("org.Hs.eg.db"))
suppressPackageStartupMessages(library("GO.db"))

Introduction

The HPA project

From the Human Protein Atlas [@Uhlen2005; @Uhlen2010] site:

The Swedish Human Protein Atlas project, funded by the Knut and Alice Wallenberg Foundation, has been set up to allow for a systematic exploration of the human proteome using Antibody-Based Proteomics. This is accomplished by combining high-throughput generation of affinity-purified antibodies with protein profiling in a multitude of tissues and cells assembled in tissue microarrays. Confocal microscopy analysis using human cell lines is performed for more detailed protein localisation. The program hosts the Human Protein Atlas portal with expression profiles of human proteins in tissues and cells.

The r Biocpkg("hpar") package provides access to HPA data from the R interface. It also distributes the following data sets:

Several flat files are distributed by the HPA project and available within the package as data.frames, other datasets are available through a search query on the HPA website. The description below is taken from the HPA site:

hpaNormalTissue: Normal tissue data. Expression profiles for proteins in human tissues based on immunohistochemisty using tissue micro arrays. The tab-separated file includes Ensembl gene identifier ("Gene"), tissue name ("Tissue"), annotated cell type ("Cell type"), expression value ("Level"), and the gene reliability of the expression value ("Reliability").
hpaNormalTissue16.1: Same as above, for version 16.1.
hpaCancer: Pathology data. Staining profiles for proteins in human tumor tissue based on immunohistochemisty using tissue micro arrays and log-rank P value for Kaplan-Meier analysis of correlation between mRNA expression level and patient survival. The tab-separated file includes Ensembl gene identifier ("Gene"), gene name ("Gene name"), tumor name ("Cancer"), the number of patients annotated for different staining levels ("High", "Medium", "Low" & "Not detected") and log-rank p values for patient survival and mRNA correlation ("prognostic - favourable", "unprognostic - favourable", "prognostic - unfavourable", "unprognostic - unfavourable").
hpaCancer16.1: Same as above, for version 16.1
rnaConsensusTissue: RNA consensus tissue gene data. Consensus transcript expression levels summarized per gene in 54 tissues based on transcriptomics data from HPA and GTEx. The consensus normalized expression ("nTPM") value is calculated as the maximum nTPM value for each gene in the two data sources. For tissues with multiple sub-tissues (brain regions, lymphoid tissues and intestine) the maximum of all sub-tissues is used for the tissue type. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Tissue") and normalized expression ("nTPM").
rnaHpaTissue: RNA HPA tissue gene data. Transcript expression levels summarized per gene in 256 tissues based on RNA-seq. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Tissue"), transcripts per million ("TPM"), protein-transcripts per million ("pTPM") and normalized expression ("nTPM").
rnaGtexTissue: RNA GTEx tissue gene data. Transcript expression levels summarized per gene in 37 tissues based on RNA-seq. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Tissue"), transcripts per million ("TPM"), protein-transcripts per million ("pTPM") and normalized expression ("nTPM"). The data was obtained from GTEx.
rnaFantomTissue: RNA FANTOM tissue gene data. Transcript expression levels summarized per gene in 60 tissues based on CAGE data. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Tissue"), tags per million ("Tags per million"), scaled-tags per million ("Scaled tags per million") and normalized expression ("nTPM"). The data was obtained from FANTOM5.
rnaGeneTissue21.0: RNA HPA tissue gene data. Transcript expression levels summarized per gene in 37 tissues based on RNA-seq, for hpa version 21.0. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Tissue"), transcripts per million ("TPM"), protein-transcripts per million ("pTPM").
rnaGeneCellLine: RNA HPA cell line gene data. Transcript expression levels summarized per gene in 69 cell lines. The tab-separated file includes Ensembl gene identifier ("Gene"), analysed sample ("Cell line"), transcripts per million ("TPM"), protein-coding transcripts per million ("pTPM") and normalized expression ("nTPM").
rnaGeneCellLine16.1: Same as above, for version 16.1.
hpaSubcellularLoc: Subcellular location data. Subcellular location of proteins based on immunofluorescently stained cells. The tab-separated file includes the following columns: Ensembl gene identifier ("Gene"), name of gene ("Gene name"), gene reliability score ("Reliability"), enhanced locations ("Enhanced"), supported locations ("Supported"), Approved locations ("Approved"), uncertain locations ("Uncertain"), locations with single-cell variation in intensity ("Single-cell variation intensity"), locations with spatial single-cell variation ("Single-cell variation spatial"), locations with observed cell cycle dependency (type can be one or more of biological definition, custom data or correlation) ("Cell cycle dependency"), Gene Ontology Cellular Component term identifier ("GO id").
hpaSubcellularLoc16.1: Same as above, for version 16.1.
hpaSubcellularLoc14: Same as above, for version 14.
hpaSecretome: Secretome data. The human secretome is here defined as all Ensembl genes with at least one predicted secreted transcript according to HPA predictions. The complete information about the HPA Secretome data is given on https://www.proteinatlas.org/humanproteome/blood/secretome. This dataset has 315 columns and includes the Ensembl gene identifier ("Gene"). Information about the additionnal variables can be found here by clicking on Show/hide columns.

The hpar::allHparData() returns a list of all datasets (see below).

HPA data usage policy

The use of data and images from the HPA in publications and presentations is permitted provided that the following conditions are met:

The publication and/or presentation are solely for informational and non-commercial purposes.
The source of the data and/or image is referred to the HPA site^[www.proteinatlas.org] and/or one or more of our publications are cited.

Installation

r Biocpkg("hpar") is available through the Bioconductor project. Details about the package and the installation procedure can be found on its landing page. To install using the dedicated Bioconductor infrastructure, run :

## install BiocManager only one
install.packages("BiocManager")
## install hpar
BiocManager::install("hpar")

After installation, r Biocpkg("hpar") will have to be explicitly loaded with

library("hpar")

so that all the package's functionality and data is available to the user.

The `r Biocpkg("hpar")` package

Data sets

A table descibing all dataset available in the package can be accessed with the allHparData() function.

hpa_data <- allHparData()

DT::datatable(hpa_data)

The Title variable corresponds to names of the data that can be downloaded localled and cached as part of the ExperimentHub infrastructure.

head(normtissue <- hpaNormalTissue())

Note that given that the hpa data is distributed as par the ExperimentHub infrastructure, it is also possible to query it directly for relevant datasets.

library("ExperimentHub")
eh <- ExperimentHub()
query(eh, "hpar")

HPA interface

Each data described above is a data.frame and can be easily manipulated using standard R or BiocStyle::CRANpkg("tidyverse") tidyverse functionality.

names(normtissue)
## Number of genes
length(unique(normtissue$Gene))
## Number of cell types
length(unique(normtissue$Cell.type))
## Number of tissues
length(unique(normtissue$Tissue))

## Number of genes highlighly and reliably expressed in each cell type
## in each tissue.
library("dplyr")
normtissue |>
    filter(Reliability == "Approved",
           Level == "High") |>
    count(Cell.type, Tissue) |>
    arrange(desc(n)) |>
    head()

We will illustrate additional datasets using the TSPAN6 (tetraspanin 6) gene (ENSG00000000003) as example.

id <- "ENSG00000000003"
subcell <- hpaSubcellularLoc()
rna <- rnaGeneCellLine()

## Compine protein immunohistochemisty data, with the subcellular
## location and RNA expression levels.
filter(normtissue, Gene == id) |>
    full_join(filter(subcell, Gene == id)) |>
    full_join(filter(rna, Gene == id)) |>
    head()

It is also possible to directly open the HPA page for a specific gene (see figure below).

browseHPA(id)

The HPA web page for the tetraspanin 6 gene (ENSG00000000003).

HPA release information

Information about the HPA release used to build the installed r Biocpkg("hpar") package can be accessed with getHpaVersion, getHpaDate and getHpaEnsembl. Full release details can be found on the HPA release history page.

getHpaVersion()
getHpaDate()
getHpaEnsembl()

A small use case

Let's compare the subcellular localisation annotation obtained from the HPA subcellular location data set and the information available in the Bioconductor annotation packages.

id <- "ENSG00000001460"
filter(subcell, Gene == id)

Below, we first extract all cellular component GO terms available for id from the r Biocannopkg("org.Hs.eg.db") human annotation and then retrieve their term definitions using the r Biocannopkg("GO.db") database.

library("org.Hs.eg.db")
library("GO.db")
ans <- AnnotationDbi::select(org.Hs.eg.db, keys = id,
                             columns = c("ENSEMBL", "GO", "ONTOLOGY"),
                             keytype = "ENSEMBL")
ans <- ans[ans$ONTOLOGY == "CC", ]
ans
sapply(as.list(GOTERM[ans$GO]), slot, "Term")