mzidentml-reader

mzidentml-reader processes mzIdentML 1.2.0 and 1.3.0 files with the primary aim of extracting crosslink information. It has three use cases:

1. to validate mzIdentML files against the criteria given here: https://www.ebi.ac.uk/pride/markdownpage/crosslinking
2. to extract information on crosslinked residue pairs and output it in a form more easily used by modelling software
3. to populate the database that is accessed by crosslinking-api

It uses the pyteomics library (https://pyteomics.readthedocs.io/en/latest/index.html) as the underlying parser for mzIdentML. Results are written into a relational database (PostgreSQL or SQLite) using sqlalchemy.

Requirements

• Python 3.10 (includes SQLite3 in standard library)
• pipenv (for dependency management)
• PostgreSQL server (optional, only required for crosslinking-api database creation; validation and residue pair extraction use built-in SQLite3)

Installation

Production Installation

Install via PyPI:

pip install mzidentml-reader

PyPI project: https://pypi.org/project/mzidentml-reader/

For more installation details, see: https://packaging.python.org/en/latest/tutorials/installing-packages/

Development Setup

Clone the repository and set up the development environment:

git clone https://github.com/PRIDE-Archive/mzidentml-reader.git
cd mzidentml-reader
pipenv install --python 3.10 --dev
pipenv shell

Usage

process_dataset is the CLI entry point. Run it with -h to see all options:

process_dataset -h

Alternative (from the repository root):

python -m parser -h

CLI Options Reference

One of the following mutually exclusive options is required:

Option	Description
-p, --pxid <ID> [ID ...]	ProteomeXchange accession(s), e.g. PXD000001 or numbers only. Multiple IDs can be space-separated.
-f, --ftp <URL>	Process files from the specified FTP location.
-d, --dir <PATH>	Process files in the specified local directory.
-v, --validate <PATH>	Validate an mzIdentML file or all files in a directory. Exits after first failure.
--seqsandresiduepairs <PATH>	Extract sequences and crosslinked residue pairs as JSON. Requires -j.

Additional options:

Option	Description	Default
-t, --temp <PATH>	Temp folder for downloaded files or the sqlite DB.	System temp directory
-n, --nopeaklist	Skip peak list file checks. Works with -d and -v only.	Off
-w, --writer <db|api>	Save data to database (db) or API (api). Used with -p, -f, -d.	db
-j, --json <FILE>	Output JSON filename. Required when using --seqsandresiduepairs.
-i, --identifier <ID>	Project identifier for the database. Defaults to PXD accession or directory name.
--dontdelete	Don't delete downloaded data after processing.	Off

1. Validate a dataset

Run with the -v option to validate a dataset. The argument is the path to a specific mzIdentML file or to a directory containing multiple mzIdentML files, in which case all of them will be validated. To pass, all the peaklist files referenced must be in the same directory as the mzIdentML file(s). The converter will create an sqlite database in the temporary folder which is used in the validation process, the temporary folder can be specified with the -t option.

Use -n to skip peak list file checks (useful when peak list files are not available locally):

Examples:

process_dataset -v ~/mydata

process_dataset -v ~/mydata/mymzid.mzid -t ~/mytempdir

process_dataset -v ~/mydata/mymzid.mzid -n

The result is written to the console. If the data fails validation but the error message is not informative, please open an issue on the github repository: https://github.com/PRIDE-Archive/mzidentml-reader/issues

2. Extract summary of crosslinked residue pairs

Run with the --seqsandresiduepairs option to extract a summary of search sequences and crosslinked residue pairs. The output is JSON which is written to a file specified with the -j option (required). The argument is the path to an mzIdentML file or a directory containing multiple mzIdentML files, in which case all of them will be processed.

Examples:

process_dataset --seqsandresiduepairs ~/mydata -j output.json -t ~/mytempdir

process_dataset --seqsandresiduepairs ~/mydata/mymzid.mzid -j output.json

Programmatic access

The functionality can also be accessed programmatically in Python:

from parser.process_dataset import sequences_and_residue_pairs
import tempfile

# Get sequences and residue pairs as a dictionary
filepath = "/path/to/file.mzid"  # or directory containing .mzid files
tmpdir = tempfile.gettempdir()   # or specify your own temp directory

data = sequences_and_residue_pairs(filepath, tmpdir)

# Iterate through sequences
print(f"Found {len(data['sequences'])} sequences:")
for seq in data['sequences']:
    print(f"  {seq['accession']}: {seq['sequence'][:50]}... (from {seq['file']})")

# Iterate through crosslinked residue pairs
print(f"\nFound {len(data['residue_pairs'])} unique crosslinked residue pairs:")
for pair in data['residue_pairs']:
    print(f"  {pair['prot1_acc']}:{pair['pos1']} <-> {pair['prot2_acc']}:{pair['pos2']}")
    print(f"    Match IDs: {pair['match_ids']}")
    print(f"    Modification accessions: {pair['mod_accs']}")

The returned dictionary has two keys:

• sequences: List of protein sequences (id, file, sequence, accession)
• residue_pairs: List of crosslinked residue pairs (prot1, prot1_acc, pos1, prot2, prot2_acc, pos2, match_ids, files, mod_accs)

3. Populate the crosslinking-api database

Create the database

sudo su postgres;
psql;
create database crosslinking;
create user xiadmin with login password 'your_password_here';
grant all privileges on database crosslinking to xiadmin;
\connect crosslinking;
GRANT ALL PRIVILEGES ON SCHEMA public TO xiadmin;

find the hba.conf file in the postgresql installation directory and add a line to allow the xiadmin role to access the database: e.g.

sudo nano /etc/postgresql/13/main/pg_hba.conf

then add the line: local crosslinking xiadmin md5

then restart postgresql:

sudo service postgresql restart

Configure the python environment for the file parser

edit the file mzidentml-reader/config/database.ini to point to your postgressql database. e.g. so its content is:

[postgresql]
host=localhost
database=crosslinking
user=xiadmin
password=your_password_here
port=5432

Create the database schema

run create_db_schema.py to create the database tables:

python parser/database/create_db_schema.py

Populate the database

To parse a test dataset:

process_dataset -d ~/PXD038060

The command line options that populate the database are -d, -f and -p. Only one of these can be used.

• -d — process files in a local directory
• -f — process files from an FTP location
• -p — process by ProteomeXchange identifier(s), space-separated

The -i option sets the project identifier in the database. It defaults to the PXD accession or the name of the directory containing the mzIdentML file.

The -w option selects the writer method (db for database, api for API). Defaults to db.

Use --dontdelete to keep downloaded data after processing.

Examples:

process_dataset -p PXD038060

process_dataset -p PXD038060 PXD000001 -w api

process_dataset -f ftp://ftp.jpostdb.org/JPST001914/ -i JPST001914

4. Cleanup noncov modifications

The cleanup_noncov module removes invalid crosslink donor/acceptor modifications (location="-1") from mzIdentML files. This is useful for pre-processing files that contain noncovalent modifications that are not properly located.

Programmatic access

from parser.cleanup_noncov import cleanup_noncov, cleanup_noncov_gz

# For plain .mzid files
peps_cleaned, mods_removed, sii_cleaned = cleanup_noncov("input.mzid", "output.mzid")

# For gzipped .mzid.gz files
peps_cleaned, mods_removed, sii_cleaned = cleanup_noncov_gz("input.mzid.gz", "output.mzid.gz")

print(f"Peptides cleaned: {peps_cleaned}")
print(f"Modifications removed: {mods_removed}")
print(f"SpectrumIdentificationItems cleaned: {sii_cleaned}")

Development

Code Quality

This project uses standardized code quality tools:

# Format code
pipenv run black .

# Sort imports
pipenv run isort .

# Check style and syntax
pipenv run flake8

Testing

Make sure the test database user is available:

psql -p 5432 -c "create role ximzid_unittests with password 'ximzid_unittests';"
psql -p 5432 -c 'alter role ximzid_unittests with login;'
psql -p 5432 -c 'alter role ximzid_unittests with createdb;'
psql -p 5432 -c 'GRANT pg_signal_backend TO ximzid_unittests;'

Run tests with coverage:

pipenv run pytest  # Run tests with coverage (80% threshold)
pipenv run pytest --cov-report=html  # Generate HTML coverage report
pipenv run pytest -m "not slow"  # Skip slow tests