#xml-parser #website #csv #downloaded #uni-parc #uniprot #uniparc

bin+lib biorustlings

Scripts for parsing UniParc XML files downloaded from the Uniprot website into CSV files

1 unstable release

Uses old Rust 2015

0.0.2 Nov 13, 2017

#12 in #downloaded

MIT license

47KB
1K SLoC

Rust 592 SLoC // 0.0% comments Python 535 SLoC // 0.1% comments Forge Config 28 SLoC

UniParc XML parser

Process the UniParc XML file (uniparc_all.xml.gz) downloaded from the UniProt website into CSV files that can be loaded into a relational database.

Example

Parsing 1 million lines takes about 5.5 seconds:

$ mkdir uniparc
$ time bash -c "zcat tests/uniparc_1mil.xml.gz | uniparc_xml_parser >/dev/null"

real    0m5.564s
user    0m5.528s
sys     0m0.132s

The actual uniparc_all.xml.gz file is about 5 billion rows.

FAQ

Why not split uniparc_all.xml.gz into multiple small files and process them in parallel

  • Splitting the file requires reading the entire file. If we're reading the entire file anyway, why not parse it as we read it?
  • Having a single process which parses uniparc_all.xml.gz makes it easier to create an incremental unique index column (e.g. UniparcXRef.idx, Property.idx, etc.).

Dependencies

~400KB