proteomics-analysis

Proteomics Analysis

Overview

Mass spectrometry-based proteomics enables global identification and quantification of proteins in complex biological samples. This skill covers data processing from MaxQuant/DIA-NN output files, statistical analysis, differential expression, post-translational modification (PTM) analysis, and visualization of proteomic data.

When to Use This Skill

• Processing MaxQuant proteinGroups.txt or peptides.txt output files
• Analyzing DIA-NN report files for data-independent acquisition experiments
• Statistical differential expression analysis between conditions
• Label-free quantification (LFQ), TMT, and SILAC analysis
• PTM analysis (phosphoproteomics, ubiquitination, acetylation)
• Protein-protein interaction network analysis from AP-MS data
• Visualizing protein abundance, volcano plots, and heatmaps
• Integration with transcriptomics data (multi-omics)

Quick Start

Processing MaxQuant Output

import pandas as pd
import numpy as np

# Load MaxQuant proteinGroups
pg = pd.read_csv("proteinGroups.txt", sep="\t", low_memory=False)

# Basic filtering
pg_filtered = pg[
    (pg["Reverse"] != "+") &
    (pg["Potential contaminant"] != "+") &
    (pg["Only identified by site"] != "+")
].copy()

print(f"Proteins before filter: {len(pg)}")
print(f"Proteins after filter: {len(pg_filtered)}")

# Extract LFQ intensity columns
lfq_cols = [c for c in pg_filtered.columns if c.startswith("LFQ intensity")]
print(f"Samples: {len(lfq_cols)}")
print(lfq_cols)

# Replace 0 with NaN (missing values)
intensity_matrix = pg_filtered[lfq_cols].replace(0, np.nan)
intensity_matrix.index = pg_filtered["Gene names"].fillna(pg_filtered["Protein IDs"])

# Log2 transform
log2_matrix = np.log2(intensity_matrix)
print(f"\nLog2 intensity range: {log2_matrix.min().min():.1f} - {log2_matrix.max().max():.1f}")

Missing Value Imputation

import pandas as pd
import numpy as np
from sklearn.impute import KNNImputer

def impute_missing_values(df: pd.DataFrame, method: str = "knn") -> pd.DataFrame:
    """Impute missing values in proteomics matrix."""
    if method == "knn":
        imputer = KNNImputer(n_neighbors=5)
        imputed = imputer.fit_transform(df.T)
        return pd.DataFrame(imputed.T, index=df.index, columns=df.columns)
    elif method == "min_based":
        # MinProb: impute from left tail of distribution (for MNAR)
        result = df.copy()
        for col in df.columns:
            col_min = df[col].quantile(0.01)
            col_std = df[col].std() * 0.3
            n_missing = df[col].isna().sum()
            result.loc[df[col].isna(), col] = np.random.normal(
                col_min, col_std, n_missing
            )
        return result
    else:
        return df.fillna(df.median())

# Apply imputation
log2_imputed = impute_missing_values(log2_matrix, method="knn")