Bio Proteomics Data Import Agent Skill

Version Compatibility

Reference examples tested with: MSnbase 2.28+, pandas 2.2+

Before using code patterns, verify installed versions match. If versions differ:

• Python: pip show <package> then help(module.function) to check signatures
• R: packageVersion('<pkg>') then ?function_name to verify parameters

If code throws ImportError, AttributeError, or TypeError, introspect the installed

package and adapt the example to match the actual API rather than retrying.

Mass Spectrometry Data Import

"Load my mass spec data into Python" → Parse mzML/mzXML raw files or MaxQuant proteinGroups.txt into data structures for programmatic access and downstream analysis.

• Python: pyopenms.MzMLFile().load() for raw spectra, pandas.read_csv() for search engine outputs
• R: MSnbase::readMSData() for raw, read.delim() for MaxQuant/Proteome Discoverer

Loading mzML/mzXML Files with pyOpenMS

Goal: Parse raw mass spectrometry data files into memory for programmatic access.

Approach: Load mzML/mzXML into an MSExperiment object, then iterate spectra by MS level to access peaks and precursor info.

from pyopenms import MSExperiment, MzMLFile, MzXMLFile

exp = MSExperiment()
MzMLFile().load('sample.mzML', exp)

for spectrum in exp:
    if spectrum.getMSLevel() == 1:
        mz, intensity = spectrum.get_peaks()
    elif spectrum.getMSLevel() == 2:
        precursor = spectrum.getPrecursors()[0]
        precursor_mz = precursor.getMZ()

Loading MaxQuant Output

Goal: Import MaxQuant proteinGroups.txt with contaminant and decoy filtering.

Approach: Read the TSV file, remove reverse hits, contaminants, and site-only identifications, then extract intensity columns.

import pandas as pd

protein_groups = pd.read_csv('proteinGroups.txt', sep='\t', low_memory=False)

# Filter contaminants and reverse hits
contam_col = 'Potential contaminant' if 'Potential contaminant' in protein_groups.columns else 'Contaminant'
protein_groups = protein_groups[
    (protein_groups.get(contam_col, '') != '+') &
    (protein_groups.get('Reverse', '') != '+') &
    (protein_groups.get('Only identified by site', '') != '+')
]

# Extract intensity columns (LFQ or iBAQ)
intensity_cols = [c for c in protein_groups.columns if c.startswith('LFQ intensity') or c.startswith('iBAQ ')]
if not intensity_cols:
    intensity_cols = [c for c in protein_groups.columns if c.startswith('Intensity ') and 'Intensity L' not in c]
intensities = protein_groups[['Protein IDs', 'Gene names'] + intensity_cols]

Loading Spectronaut/DIA-NN Output

Goal: Import DIA-NN long-format report and reshape into a protein-by-sample quantification matrix.

Approach: Pivot the report table on protein group and run columns, using MaxLFQ values.

diann_report = pd.read_csv('report.tsv', sep='\t')

# Pivot to protein-level matrix
protein_matrix = diann_report.pivot_table(
    index='Protein.Group', columns='Run', values='PG.MaxLFQ', aggfunc='first'
)

R: Loading with MSnbase

Goal: Load raw MS data in R for interactive exploration of spectra and metadata.

Approach: Use MSnbase's on-disk reading mode to access spectra and feature metadata without loading all data into memory.

library(MSnbase)

raw_data <- readMSData('sample.mzML', mode = 'onDisk')
spectra <- spectra(raw_data)
header_info <- fData(raw_data)

Missing Value Assessment

Goal: Quantify missing value patterns across proteins and samples in an intensity matrix.

Approach: Count NaN values per protein and per sample, then compute overall missing percentage.

def assess_missing_values(df, intensity_cols):
    missing_per_protein = df[intensity_cols].isna().sum(axis=1)
    missing_per_sample = df[intensity_cols].isna().sum(axis=0)

    total_missing = df[intensity_cols].isna().sum().sum()
    total_values = df[intensity_cols].size
    missing_pct = 100 * total_missing / total_values

    return {'per_protein': missing_per_protein, 'per_sample': missing_per_sample, 'total_pct': missing_pct}

Related Skills

• quantification - Process imported data for quantification
• peptide-identification - Identify peptides from raw spectra
• expression-matrix/counts-ingest - Similar data loading patterns