4. Tutorial: Working with Human Protein Atlas (HPA) xml files offline

library(BiocStyle)
library(HPAanalyze)
library(dplyr)
library(xml2)

The case

The Human Protein Atlas allow you to download very detailed data for each protein in the form of an xml file, and hpaXmlGet and hpaXml allow you to retrieve those files automatically from HPA server and parse them. However, due to technical limitation, you will not be able to save those "xml_document"/"xml_node" objects. The question is: How do you keep a version of these files to use when you are not connected to the internet, or for reproducibility?

The solution

Download and keep a local version of the xml files for yourself

Look at the “Downloadable data” page from HPA website, you will see how these files are downloaded. Basically, you add [ensembl_id].xml to http://www.proteinatlas.org to download individual entries (that’s what hpaXmlGet does behind the scene), or download the whole big set.

From there, you can import the file using xml2::read_xml(). The output should be exactly the same as hpaXmlGet.

## same as hpaXmlGet("ENSG00000134057")
CCNB1xml <- xml2::read_xml("data/ENSG00000134057.xml")

Business as usual with hpaXml functions

Since the umbrella function hpaXml take either the ensembl id or the imported xml_document object, you can feed what you just imported to it and get the expected result.

CCNB1_parsed <- hpaXml(CCNB1xml)

You can obviously use other hpaXml functions as well.

hpaXmlProtClass(CCNB1xml)
hpaXmlTissueExprSum(CCNB1xml)
hpaXmlAntibody(CCNB1xml)
hpaXmlTissueExpr(CCNB1xml)

Save your parsed objects

It is recommended that you save your parsed objects for reproducibility. Unlike the xml_document object, these parsed objects are just regular R lists of standard vectors or data frames. You can save them just as usual.

saveRDS(CCNB1_parsed, "data/CCNB1_parsed.rds")