10.1 Protein structure databases

Protein structure databases are essential tools in bioinformatics, providing researchers with vast repositories of 3D protein structures. These databases enable scientists to analyze protein function, evolution, and interactions, supporting various applications from drug design to evolutionary studies.

Understanding the types, formats, and search methods of protein structure databases is crucial for bioinformaticians. By leveraging these resources effectively, researchers can gain valuable insights into protein behavior and develop innovative solutions to biological problems.

Types of protein databases

• Protein databases serve as essential resources in bioinformatics, providing researchers with vast repositories of protein information
• These databases play a crucial role in advancing our understanding of protein structure, function, and evolution
• Bioinformaticians utilize various types of protein databases to analyze and interpret complex biological data

Primary vs derivative databases

Top images from around the web for Primary vs derivative databases

• 

  Frontiers | Grand Challenges in Bioinformatics Data Visualization View original

  Is this image relevant?

• 

  Frontiers | An Improved Sequencing-Based Bioinformatics Pipeline to Track the Distribution and ... View original

  Is this image relevant?

• 

  Frontiers | Bioinformatic Analysis of Temporal and Spatial Proteome Alternations During Infections View original

  Is this image relevant?

• 

  Frontiers | Grand Challenges in Bioinformatics Data Visualization View original

  Is this image relevant?

• 

  Frontiers | An Improved Sequencing-Based Bioinformatics Pipeline to Track the Distribution and ... View original

  Is this image relevant?

1 of 3

Top images from around the web for Primary vs derivative databases

• 

  Frontiers | Grand Challenges in Bioinformatics Data Visualization View original

  Is this image relevant?

• 

  Frontiers | An Improved Sequencing-Based Bioinformatics Pipeline to Track the Distribution and ... View original

  Is this image relevant?

• 

  Frontiers | Bioinformatic Analysis of Temporal and Spatial Proteome Alternations During Infections View original

  Is this image relevant?

• 

  Frontiers | Grand Challenges in Bioinformatics Data Visualization View original

  Is this image relevant?

• 

  Frontiers | An Improved Sequencing-Based Bioinformatics Pipeline to Track the Distribution and ... View original

  Is this image relevant?

1 of 3

• Primary databases contain experimentally determined data directly submitted by researchers
• Derivative databases compile and curate information from primary databases, often adding value through annotations and analyses
• Primary databases (GenBank) focus on raw sequence or structure data
• Derivative databases (UniProtKB) offer additional layers of information, including functional annotations and cross-references

Sequence vs structure databases

• Sequence databases store protein amino acid sequences, enabling researchers to analyze primary structures
• Structure databases contain three-dimensional protein structures determined through experimental methods (X-ray crystallography, NMR spectroscopy)
• Sequence databases (UniProtKB) facilitate sequence alignment, homology detection, and evolutionary studies
• Structure databases (PDB) support structural analysis, protein folding research, and drug design efforts

Major protein structure databases

• Protein structure databases form the backbone of structural bioinformatics research and applications
• These databases provide researchers with access to experimentally determined three-dimensional protein structures
• Bioinformaticians leverage these resources for various tasks, including structure prediction, drug design, and evolutionary studies

Protein Data Bank (PDB)

• Centralized repository for experimentally determined 3D structures of biological macromolecules
• Contains structures of proteins, nucleic acids, and complex assemblies
• Provides standardized data formats (PDB, mmCIF) for structure representation
• Offers tools for structure visualization, analysis, and validation
• Regularly updated with new structures submitted by researchers worldwide

UniProt and SwissProt

• UniProt serves as a comprehensive protein sequence and functional information database
• SwissProt represents a manually curated subset of UniProt with high-quality annotations
• UniProt integrates data from various sources, including sequence databases and literature
• Provides extensive cross-references to other databases and resources
• Offers tools for sequence analysis, including multiple sequence alignment and domain prediction

SCOP and CATH

• SCOP (Structural Classification of Proteins) organizes protein structures based on evolutionary relationships
• CATH (Class, Architecture, Topology, Homologous superfamily) classifies protein structures hierarchically
• Both databases facilitate the study of protein evolution and structure-function relationships
• SCOP uses a manual curation process to classify structures into families and superfamilies
• CATH employs a combination of automated and manual methods for structure classification

Data representation formats

• Standardized data formats enable efficient storage, exchange, and analysis of protein structure information
• These formats capture various aspects of protein structures, including atomic coordinates and metadata
• Bioinformaticians must be familiar with different formats to effectively work with structural data

PDB file format

• Text-based format developed by the Protein Data Bank for representing 3D structures
• Contains atomic coordinates, experimental details, and metadata
• Organized into records with fixed column widths for different types of information
• Includes ATOM records for atomic coordinates and HETATM records for non-standard residues
• Supports representation of multiple models (NMR structures) and biological assemblies

mmCIF format

• Macromolecular Crystallographic Information File format, an extension of the CIF standard
• Addresses limitations of the PDB format, such as file size restrictions and limited metadata
• Uses a flexible key-value pair system to represent structural and experimental information
• Supports more detailed descriptions of experimental methods and structure quality
• Allows for easier parsing and automated processing of structural data

XML-based formats

• XML (eXtensible Markup Language) formats provide a hierarchical representation of protein structure data
• PDBML (Protein Data Bank Markup Language) represents PDB data in XML format
• mmCIF2XML converts mmCIF data into XML format for improved interoperability
• XML-based formats facilitate data exchange and integration with other bioinformatics tools
• Enable easier parsing and validation of structural data using standard XML tools

Database search methods

• Efficient search methods allow researchers to retrieve relevant protein structure information from databases
• Various search strategies cater to different research needs and data types
• Bioinformaticians employ these search methods to identify structures of interest for further analysis

Sequence-based searches

• BLAST (Basic Local Alignment Search Tool) identifies similar sequences in protein databases
• PSI-BLAST (Position-Specific Iterative BLAST) performs iterative searches for distant homologs
• Sequence motif searches identify specific patterns or domains within protein sequences
• Multiple sequence alignment tools (Clustal Omega) compare and align related protein sequences
• Profile Hidden Markov Models (HMMs) detect remote homologs based on sequence patterns

Structure-based searches

• DALI (Distance matrix ALIgnment) compares protein structures based on distance matrices
• CE (Combinatorial Extension) aligns protein structures using secondary structure elements
• VAST (Vector Alignment Search Tool) performs rapid structure similarity searches
• Structural motif searches identify specific 3D arrangements of amino acids or secondary structures
• Ligand-based searches find structures containing similar binding sites or bound molecules

Keyword and metadata searches

• Text-based searches allow users to find structures based on protein names, functions, or organisms
• Advanced search options combine multiple criteria (resolution, experimental method, publication date)
• Ontology-based searches utilize standardized vocabularies (Gene Ontology) for consistent annotations
• Author name searches retrieve structures associated with specific researchers or laboratories
• Literature-based searches find structures mentioned in scientific publications

Data quality and validation

• Ensuring the quality and reliability of protein structure data is crucial for accurate analysis and interpretation
• Various metrics and tools help assess the quality of experimentally determined structures
• Bioinformaticians must consider data quality when selecting structures for analysis or modeling

Experimental methods in structures

• X-ray crystallography determines atomic positions by analyzing X-ray diffraction patterns
• Nuclear Magnetic Resonance (NMR) spectroscopy measures distances between atoms in solution
• Cryo-electron microscopy (cryo-EM) visualizes macromolecular structures at near-atomic resolution
• Each method has strengths and limitations in terms of resolution, sample preparation, and structure size
• Understanding experimental methods helps interpret structural data and assess its reliability

Resolution and R-factor

• Resolution measures the level of detail in an X-ray crystallography or cryo-EM structure
• Lower resolution values (1-2 Å) indicate higher-quality structures with more precise atomic positions
• R-factor quantifies the agreement between the experimental data and the refined structural model
• Lower R-factors (<0.2) suggest better agreement between the model and experimental data
• Free R-factor (R-free) provides an unbiased estimate of model quality using a test set of reflections

Structure validation tools

• MolProbity assesses the overall quality of protein structures using various geometric criteria
• PROCHECK evaluates the stereochemical quality of protein structures
• WHAT_CHECK performs extensive checks on protein structure quality and identifies potential errors
• Ramachandran plots visualize the distribution of backbone dihedral angles in protein structures
• B-factor analysis examines the thermal motion or uncertainty of atoms in crystal structures

Integration with other resources

• Integration of protein structure databases with other biological resources enhances their utility
• Cross-referencing and data integration enable researchers to connect structural information with other types of biological data
• Bioinformaticians leverage these integrated resources to gain comprehensive insights into protein function and behavior

Cross-references to other databases

• UniProt provides extensive cross-references to various biological databases
• Gene Ontology (GO) terms link protein structures to standardized functional annotations
• Enzyme Commission (EC) numbers connect structures to specific enzymatic activities
• Pfam links structures to protein domain families and their functional annotations
• KEGG (Kyoto Encyclopedia of Genes and Genomes) maps structures to metabolic pathways

Pathway and interaction databases

• STRING database integrates protein-protein interaction data with structural information
• Reactome links protein structures to biological pathways and reactions
• IntAct provides detailed information on molecular interactions involving structured proteins
• BioCyc connects protein structures to metabolic pathways and regulatory networks
• PDBe-KB (Protein Data Bank in Europe - Knowledge Base) aggregates annotations and predictions for PDB structures

Visualization tools

• PyMOL offers advanced 3D visualization and analysis of protein structures
• Chimera provides a user-friendly interface for structure visualization and manipulation
• Jmol enables web-based 3D visualization of protein structures
• NGL Viewer allows for interactive visualization of large macromolecular complexes
• Mol* Viewer integrates with the PDB website for seamless structure exploration

Programmatic access

• Programmatic access to protein structure databases enables automated data retrieval and analysis
• Various tools and interfaces allow bioinformaticians to integrate structural data into custom workflows
• These methods facilitate large-scale analyses and the development of specialized bioinformatics tools

RESTful APIs

• PDB provides a RESTful API for querying and retrieving structural data
• UniProt offers a comprehensive API for accessing protein sequence and functional information
• RCSB PDB Web Services enable programmatic access to various search and analysis tools
• PDBe REST API allows retrieval of structural data and annotations from the European PDB
• APIs support various output formats (JSON, XML) for easy integration with bioinformatics pipelines

Bulk data download

• FTP servers provide access to complete datasets from protein structure databases
• RCSB PDB offers weekly updates of the entire PDB archive for bulk download
• UniProt provides downloadable datasets of protein sequences and annotations
• SCOP and CATH offer downloadable classification data for offline analysis
• Bulk downloads enable local storage and processing of large structural datasets

Programmatic queries

• Biopython library provides tools for programmatic access to PDB and other structural databases
• BioPandas facilitates working with PDB files using pandas DataFrames
• PyMOL API allows for scripted analysis and visualization of protein structures
• PDB-tools offers a collection of Python scripts for manipulating PDB files
• DSSP (Define Secondary Structure of Proteins) algorithm can be integrated into custom scripts for secondary structure assignment

Applications in bioinformatics

• Protein structure databases play a crucial role in various bioinformatics applications
• These resources enable researchers to gain insights into protein function, evolution, and disease mechanisms
• Bioinformaticians leverage structural data to develop predictive models and design novel therapeutic strategies

Structure prediction

• Homology modeling uses known structures as templates to predict structures of related proteins
• Ab initio methods predict protein structures from sequence information alone
• Machine learning approaches (AlphaFold) have revolutionized protein structure prediction
• Protein structure prediction aids in understanding protein function and designing experiments
• Predicted structures serve as starting points for molecular dynamics simulations and docking studies

Drug design

• Structure-based drug design utilizes protein structures to identify potential binding sites
• Virtual screening employs structural information to screen large compound libraries
• Fragment-based drug discovery uses structural data to guide the design of novel ligands
• Protein-protein interaction inhibitors can be designed based on structural information
• Structure-guided optimization of lead compounds improves drug potency and selectivity

Evolutionary studies

• Structural alignments reveal evolutionary relationships between distantly related proteins
• Analysis of protein domains and their arrangements provides insights into protein evolution
• Structural phylogenetics incorporates 3D structure information into evolutionary tree construction
• Ancestral sequence reconstruction benefits from structural information to guide sequence predictions
• Comparative structural analysis helps identify functionally important residues conserved across species

Challenges and limitations

• Despite their immense value, protein structure databases face several challenges and limitations
• Understanding these issues is crucial for bioinformaticians to interpret and use structural data appropriately
• Ongoing efforts aim to address these challenges and improve the quality and coverage of structural data

Data redundancy

• Many protein structures in databases represent highly similar or identical proteins
• Redundancy can bias statistical analyses and machine learning models
• Clustering algorithms group similar structures to create non-redundant datasets
• PDB provides pre-computed sequence clusters at various identity thresholds
• Bioinformaticians must carefully consider redundancy when selecting datasets for analysis

Experimental bias

• Certain proteins are overrepresented in structural databases due to experimental feasibility
• Membrane proteins and large complexes are underrepresented due to technical challenges
• Structural genomics initiatives aim to address biases by targeting underrepresented protein families
• Experimental conditions (crystal packing, solution environment) may influence observed structures
• Bioinformaticians should consider potential biases when drawing conclusions from structural data

Missing or incomplete data

• Many protein structures contain unresolved regions due to flexibility or experimental limitations
• Side chain conformations may be uncertain in lower-resolution structures
• Some structures lack important ligands or cofactors present in the native state
• Experimental artifacts (truncations, mutations) may alter the observed structure
• Bioinformaticians must account for missing data when analyzing structures or building models