Proteomics Analysis

Process and analyze mass spectrometry data for protein identification and quantification.

MaxQuant Output Processing

import pandas as pd
import numpy as np

# Load MaxQuant output
protein_groups = pd.read_csv('proteinGroups.txt', sep='\t')

# Filter contaminants and reverse hits
protein_groups = protein_groups[
    ~protein_groups['Potential contaminant'].fillna('').str.contains('+') &
    ~protein_groups['Reverse'].fillna('').str.contains('+') &
    ~protein_groups['Only identified by site'].fillna('').str.contains('+')
]

# Extract LFQ intensities
lfq_cols = [col for col in protein_groups.columns if col.startswith('LFQ intensity')]
lfq_data = protein_groups[['Protein IDs', 'Gene names'] + lfq_cols]

# Log2 transform
lfq_data[lfq_cols] = np.log2(lfq_data[lfq_cols].replace(0, np.nan))

Differential Expression with limma

library(limma)

# Load data
protein_data <- read.csv("protein_intensities.csv", row.names = 1)

# Design matrix
group <- factor(c("control", "control", "control", "treated", "treated", "treated"))
design <- model.matrix(~0 + group)
colnames(design) <- levels(group)

# Fit model
fit <- lmFit(protein_data, design)

# Contrasts
contrast_matrix <- makeContrasts(treated - control, levels = design)
fit2 <- contrasts.fit(fit, contrast_matrix)
fit2 <- eBayes(fit2)

# Results
results <- topTable(fit2, adjust.method = "BH", number = Inf)

Peptide Identification with pyOpenMS

from pyopenms import *

# Load spectrum file
exp = MSExperiment()
MzMLFile().load("sample.mzML", exp)

# Access spectra
for spectrum in exp:
    if spectrum.getMSLevel() == 2:
        mz, intensity = spectrum.get_peaks()
        precursor_mz = spectrum.getPrecursors()[0].getMZ()
        print(f"Precursor: {precursor_mz:.4f}, Peaks: {len(mz)}")

# Database search setup
fasta_db = FASTAFile()
protein_ids = []
fasta_db.load("database.fasta", protein_ids)

TMT/iTRAQ Quantification

def extract_tmt_intensities(spectrum, reporter_masses, tolerance=0.01):
    """Extract TMT reporter ion intensities from MS2 spectrum."""
    mz, intensity = spectrum.get_peaks()
    intensities = {}

    for label, mass in reporter_masses.items():
        idx = np.where(np.abs(mz - mass) < tolerance)[0]
        if len(idx) > 0:
            intensities[label] = intensity[idx].sum()
        else:
            intensities[label] = 0

    return intensities

# TMT10plex reporter masses
tmt10_masses = {
    '126': 126.127726, '127N': 127.124761, '127C': 127.131081,
    '128N': 128.128116, '128C': 128.134436, '129N': 129.131471,
    '129C': 129.137790, '130N': 130.134825, '130C': 130.141145,
    '131': 131.138180
}

Protein-Protein Interaction Analysis

import networkx as nx

def build_ppi_network(interactions_file):
    """Build protein-protein interaction network."""
    interactions = pd.read_csv(interactions_file)

    G = nx.Graph()
    for _, row in interactions.iterrows():
        G.add_edge(row['protein_a'], row['protein_b'],
                  weight=row['confidence_score'])

    # Network statistics
    print(f"Nodes: {G.number_of_nodes()}")
    print(f"Edges: {G.number_of_edges()}")

    # Find highly connected proteins (hubs)
    degree_dict = dict(G.degree())
    hubs = sorted(degree_dict.items(), key=lambda x: x[1], reverse=True)[:10]

    return G, hubs

# Cluster detection
from networkx.algorithms import community
communities = community.louvain_communities(G)

Visualization

import matplotlib.pyplot as plt
import seaborn as sns

# Volcano plot
def volcano_plot(results, fc_col='log2FC', pval_col='adj.P.Val'):
    plt.figure(figsize=(10, 8))
    plt.scatter(results[fc_col], -np.log10(results[pval_col]),
               alpha=0.5, s=5)

    # Highlight significant
    sig = results[(results[pval_col] < 0.05) & (abs(results[fc_col]) > 1)]
    plt.scatter(sig[fc_col], -np.log10(sig[pval_col]),
               c='red', alpha=0.5, s=5)

    plt.xlabel('Log2 Fold Change')
    plt.ylabel('-Log10 Adjusted P-value')
    plt.axhline(-np.log10(0.05), color='gray', linestyle='--')
    plt.axvline(-1, color='gray', linestyle='--')
    plt.axvline(1, color='gray', linestyle='--')
    plt.savefig('volcano_plot.png', dpi=150)

Common Tools

• MaxQuant: Label-free quantification
• Proteome Discoverer: Thermo platform
• MSFragger: Ultra-fast database search
• Perseus: Statistical analysis
• STRING: Protein interactions database