🧬Proteomics Unit 9 – Proteomics Data Analysis and Bioinformatics

Key Concepts and Terminology

• Proteomics studies the structure, function, and interactions of proteins on a large scale
• Mass spectrometry (MS) analyzes ionized molecules based on their mass-to-charge ratio (m/z)
• Peptides are short chains of amino acids that make up proteins
  • Tryptic peptides result from digesting proteins with the enzyme trypsin
• Post-translational modifications (PTMs) alter protein function and include phosphorylation and glycosylation
• Shotgun proteomics identifies proteins by digesting them into peptides and analyzing them with MS
• Targeted proteomics focuses on specific proteins or peptides of interest using selected reaction monitoring (SRM) or parallel reaction monitoring (PRM)
• Label-free quantification compares protein abundance across samples without using stable isotope labels
• Stable isotope labeling quantifies proteins by incorporating heavy isotopes (13C, 15N) into peptides

Proteomics Data Types and Formats

• Raw MS data consists of mass spectra and chromatograms stored in proprietary formats (Thermo RAW, Waters RAW)
• Peak lists contain m/z values and intensities of detected ions and are used for database searching
  • Mascot Generic Format (MGF) and mzML are common peak list formats
• Protein sequence databases (UniProt, RefSeq) provide reference sequences for protein identification
• Spectral libraries contain previously identified spectra and can be used for spectral matching
• Quantitative data includes protein and peptide abundance values across samples
• Metadata describes experimental conditions, sample preparation, and instrument settings
• Proteomics standards initiative (PSI) develops data formats for interoperability (mzIdentML, mzQuantML)

Sample Preparation and Mass Spectrometry Basics

• Sample preparation isolates proteins from biological samples and digests them into peptides
  • Protein extraction methods include cell lysis, fractionation, and affinity purification
  • Reduction and alkylation break disulfide bonds and prevent their reformation
  • Enzymatic digestion (trypsin) cleaves proteins at specific amino acid residues (lysine, arginine)
• Liquid chromatography (LC) separates peptides based on hydrophobicity before MS analysis
• Electrospray ionization (ESI) generates gas-phase ions from liquid samples for MS analysis
• Matrix-assisted laser desorption/ionization (MALDI) ionizes samples co-crystallized with a matrix using a laser
• Tandem mass spectrometry (MS/MS) fragments peptide ions to obtain sequence information
  • Collision-induced dissociation (CID) and higher-energy collisional dissociation (HCD) are common fragmentation methods

Data Processing and Quality Control

• Raw data conversion transforms proprietary formats into open formats for analysis
• Noise reduction removes low-quality spectra and improves signal-to-noise ratio
• Charge state deconvolution determines the charge states of peptide ions
• Precursor mass correction adjusts the m/z values of peptide ions based on known masses
• Peptide spectrum matching (PSM) assigns peptide sequences to MS/MS spectra
• False discovery rate (FDR) estimation controls the proportion of false positive identifications
  • Target-decoy approach appends reversed or shuffled sequences to the database
• Quality control metrics assess the reliability of protein identifications and quantification
  • Number of PSMs, unique peptides, and protein coverage indicate identification confidence
  • Coefficient of variation (CV) measures the reproducibility of quantitative measurements

Protein Identification Algorithms

• Mascot uses a probability-based scoring algorithm to match MS/MS spectra to peptide sequences
• Sequest correlates theoretical and observed spectra using cross-correlation scores (Xcorr)
• X!Tandem employs a two-stage search strategy to identify peptides and proteins
• Andromeda is a fast and accurate search engine integrated into the MaxQuant software
• Percolator improves the sensitivity and specificity of PSMs using semi-supervised learning
• Protein inference assembles identified peptides into protein groups based on shared peptides
• Protein grouping algorithms (Occam's razor, parsimony principle) resolve ambiguities in protein assembly
• Spectral library searching compares MS/MS spectra to previously identified spectra for faster identification

Quantitative Proteomics Methods

• Label-free quantification compares protein abundance across samples based on spectral counts or ion intensities
  • Spectral counting assumes that more abundant proteins generate more PSMs
  • Ion intensity-based methods (XIC, AUC) integrate peptide ion signals across LC-MS runs
• Stable isotope labeling introduces heavy isotopes into proteins or peptides for relative quantification
  • Metabolic labeling (SILAC) incorporates heavy amino acids during cell culture
  • Chemical labeling (iTRAQ, TMT) tags peptides with isobaric reagents after digestion
• Data-independent acquisition (DIA) simultaneously fragments all precursor ions within a defined m/z range
  • Sequential window acquisition of all theoretical mass spectra (SWATH-MS) is a popular DIA method
• Targeted quantification monitors specific peptides or proteins using SRM or PRM
  • SRM detects predefined precursor-fragment ion pairs called transitions
  • PRM measures all fragment ions of a targeted precursor ion

Bioinformatics Tools and Databases

• MaxQuant is a comprehensive software package for quantitative proteomics data analysis
• Perseus performs statistical analysis and data visualization for proteomics datasets
• Skyline designs and analyzes targeted proteomics experiments (SRM, PRM)
• UniProt is a curated database of protein sequences and functional annotations
• Gene Ontology (GO) provides a standardized vocabulary for describing protein functions and locations
• Kyoto Encyclopedia of Genes and Genomes (KEGG) maps proteins to biological pathways and molecular interactions
• STRING predicts and visualizes protein-protein interaction networks based on various evidence sources
• Cytoscape is a platform for integrating and visualizing complex biological networks

Data Interpretation and Biological Insights

• Differential expression analysis identifies proteins with significant abundance changes between conditions
  • Fold change and statistical tests (t-test, ANOVA) assess the significance of expression differences
• Functional enrichment analysis reveals overrepresented biological processes, pathways, or domains in a protein list
  • Gene set enrichment analysis (GSEA) tests for the enrichment of predefined gene sets
  • Overrepresentation analysis (ORA) compares the frequency of annotations between a protein list and a background set
• Pathway mapping visualizes the involvement of identified proteins in biological pathways
• Protein-protein interaction analysis infers functional relationships and complexes among identified proteins
• Data integration combines proteomics results with other omics data (transcriptomics, metabolomics) for a systems-level understanding
• Biomarker discovery identifies proteins with diagnostic, prognostic, or predictive value for a specific condition
  • Machine learning techniques (SVM, random forests) can classify samples based on protein expression profiles
• Validation experiments confirm the biological relevance of key findings using orthogonal methods (Western blot, ELISA, immunohistochemistry)