CSIR-UGC-DBT NET LIFE SCIENCES

UNIT 11: BIOINFORMATICS AND COMPUTATIONAL BIOLOGY

SECTION A: MAJOR BIOINFORMATIC RESOURCES

Structured Notes

1. Introduction to Biological Databases

Biological databases are organised digital repositories used to store, retrieve, analyse and interpret biological information.

They may contain:

• DNA sequences

• RNA sequences

• Protein sequences

• Genome assemblies

• Gene-expression data

• Protein structures

• Protein families

• Conserved domains

• Sequence motifs

• Genetic variants

• Metabolic pathways

• Molecular interactions

• Scientific annotations

Biological databases allow researchers to:

• Retrieve biological sequences

• Compare genes and proteins

• Identify conserved domains

• Predict gene or protein function

• Study gene expression

• Examine molecular structures

• Investigate genetic variation

• Perform comparative genomics

• Reconstruct evolutionary relationships

• Design biological experiments

A biological database usually contains both biological data and associated metadata.

Metadata may include:

• Organism name

• Tissue or cell type

• Experimental method

• Sequence length

• Gene name

• Protein function

• Publication reference

• Sample source

• Accession number

• Data-quality information

2. Classification of Biological Databases

Primary or Archival Databases

Primary databases contain experimentally generated or directly submitted biological data.

Examples include:

• GenBank

• European Nucleotide Archive

• DDBJ

• Sequence Read Archive

• Gene Expression Omnibus

• Protein Data Bank

A primary database may perform quality checks, but the submitted record remains closely connected to the original experimental data.

Secondary or Derived Databases

Secondary databases contain information derived through analysis, interpretation or curation of primary data.

Examples include:

• PROSITE

• Pfam

• InterPro

• Conserved Domain Database

• PRINTS

• SMART

These resources identify:

• Protein families

• Conserved domains

• Functional sites

• Sequence motifs

• Repeats

• Evolutionary relationships

Curated or Reference Databases

Curated databases improve annotation quality through expert review, computational analysis or both.

Examples include:

• UniProtKB/Swiss-Prot

• RefSeq

Curated records are generally more reliable for functional interpretation than automatically annotated records.

However, even curated annotations may change when new evidence becomes available.

Integrated or Composite Databases

Integrated databases combine information from several different resources.

Examples include:

• InterPro

• UniProt

• NCBI Entrez

An integrated database reduces the need to search several independent resources separately.

I. SEQUENCE DATABASES

3. Nucleotide Sequence Databases

Nucleotide sequence databases store DNA and RNA sequences along with their biological annotations.

Important nucleotide resources include:

• GenBank

• European Nucleotide Archive

• DDBJ

• RefSeq

• Sequence Read Archive

4. International Nucleotide Sequence Database Collaboration

The International Nucleotide Sequence Database Collaboration or INSDC consists of:

• GenBank at NCBI

• European Nucleotide Archive at EMBL-EBI

• DNA Data Bank of Japan

These three organisations regularly exchange nucleotide-sequence records.

Therefore, a sequence submitted to one INSDC partner generally becomes available through the other partners.

Important Point

GenBank, ENA and DDBJ are not three completely independent collections of public nucleotide sequences.

They are collaborating partners that exchange sequence information.

5. GenBank

GenBank is an annotated public archive of nucleotide sequences maintained by the National Center for Biotechnology Information or NCBI.

It contains sequences submitted by:

• Individual researchers

• Genome-sequencing projects

• Research institutions

• Large sequencing consortia

• High-throughput sequencing projects

A GenBank record may contain:

• Accession number

• Sequence version

• Definition

• Organism

• Taxonomic lineage

• References

• Gene features

• Coding sequences

• Exons

• Introns

• Protein translations

• Source information

• Complete nucleotide sequence

Accession Number

An accession number is a stable identifier assigned to a database record.

When the biological sequence of a record changes, its version suffix may change.

The accession identifies the record, whereas the version identifies a particular sequence version.

GenBank Features

The feature table of a GenBank record may describe:

• Gene

• Coding sequence

• mRNA

• Exon

• Intron

• Promoter

• Regulatory region

• Repeat region

• Protein translation

• Sequence variation

Important Limitation

GenBank is an archival database containing submitted sequences.

The presence of an annotation in GenBank does not automatically mean that the annotation has been experimentally verified or manually reviewed.

6. European Nucleotide Archive

The European Nucleotide Archive or ENA is maintained by EMBL-EBI.

It provides access to:

• Raw sequencing reads

• Sequence assemblies

• Functional sequence annotations

• Genome data

• Transcriptome data

• Submitted nucleotide sequences

ENA is one of the three INSDC partners.

7. DNA Data Bank of Japan

The DNA Data Bank of Japan or DDBJ is the Japanese partner of INSDC.

It collects and distributes nucleotide-sequence data and exchanges records with GenBank and ENA.

Quick Comparison

8. RefSeq

RefSeq stands for Reference Sequence.

It is maintained by NCBI and provides an integrated, non-redundant and well-annotated collection of reference sequences.

RefSeq contains:

• Genomic sequences

• RNA sequences

• Protein sequences

• Complete chromosomes

• Organellar genomes

• Reference transcripts

• Reference proteins

RefSeq records may be generated through:

• Manual curation

• Computational annotation

• Propagation from annotated genomes

• Selection of representative sequences

Common RefSeq Prefixes

GenBank vs RefSeq

Critical Distinction

GenBank is a public sequence archive.

RefSeq is a reference sequence collection.

They should not be treated as identical resources.

9. Sequence Read Archive

The Sequence Read Archive or SRA stores raw or minimally processed high-throughput sequencing data.

It may contain data generated through:

• Whole-genome sequencing

• RNA sequencing

• ChIP sequencing

• Amplicon sequencing

• Metagenomic sequencing

• Single-cell sequencing

• Long-read sequencing

• Epigenomic sequencing

SRA data are useful for:

• Reanalysing published experiments

• Testing new bioinformatics methods

• Comparing sequencing pipelines

• Reproducing published studies

• Performing meta-analysis

SRA Data Hierarchy

Common SRA accession prefixes include:

• SRP for study

• SRS for sample

• SRX for experiment

• SRR for sequencing run

SRA vs GenBank

GenBank primarily stores assembled or annotated nucleotide sequences.

SRA primarily stores raw high-throughput sequencing reads and associated information.

10. Protein Sequence Databases

Important protein-sequence resources include:

• UniProtKB

• UniProtKB/Swiss-Prot

• UniProtKB/TrEMBL

• UniRef

• UniParc

• NCBI Protein

• RefSeq Protein

11. UniProt

UniProt stands for Universal Protein Resource.

It provides protein sequences and functional information.

The major UniProt components include:

• UniProt Knowledgebase

• UniProt Reference Clusters

• UniProt Archive

12. UniProt Knowledgebase

UniProtKB is the main protein-sequence and functional-annotation knowledgebase.

It contains two major sections:

UniProtKB/Swiss-Prot

Swiss-Prot contains reviewed protein records.

Important features include:

• Manual expert curation

• High-quality functional annotation

• Reduced redundancy

• Literature-based information

• Enzyme function

• Protein domains

• Post-translational modifications

• Subcellular location

• Disease associations

• Protein interactions

• Sequence variants

UniProtKB/TrEMBL

TrEMBL contains unreviewed protein records.

Important features include:

• Predominantly computational annotation

• Large sequence coverage

• Rapid incorporation of newly available sequences

• Limited manual review compared with Swiss-Prot

Swiss-Prot vs TrEMBL

Critical Distinction

An unreviewed UniProt record is not necessarily an incorrect protein sequence.

It means that the complete record has not undergone the same level of manual expert review as a Swiss-Prot record.

13. UniRef

UniRef groups related UniProt sequences into sequence clusters.

Its main purposes are to:

• Reduce sequence redundancy

• Increase similarity-search speed

• Simplify the analysis of large protein datasets

UniRef provides clusters at different sequence-identity thresholds.

A cluster is represented by a selected representative sequence.

14. UniParc

UniParc stands for UniProt Archive.

It stores unique protein sequences obtained from multiple source databases.

Important features include:

• Each unique sequence is stored once.

• It tracks sequence presence across source databases.

• It preserves sequence history.

• It is not a fully curated protein-annotation knowledgebase.

UniProt Resource Comparison

II. GENE-EXPRESSION DATABASES

15. Gene-Expression Data

Gene-expression experiments measure RNA abundance or transcriptional activity across:

• Tissues

• Cell types

• Developmental stages

• Disease conditions

• Environmental treatments

• Mutant and wild-type organisms

• Drug-treated and untreated samples

Common experimental methods include:

• Microarrays

• Bulk RNA sequencing

• Single-cell RNA sequencing

• Serial analysis of gene expression

• Spatial transcriptomics

Gene-expression databases store raw data, processed data and experimental metadata.

16. Gene Expression Omnibus

The Gene Expression Omnibus or GEO is maintained by NCBI.

It is a public functional-genomics repository that accepts array-based and sequence-based data.

GEO may contain:

• Microarray experiments

• RNA-seq experiments

• ChIP-seq experiments

• Methylation studies

• Expression profiling

• Epigenomic experiments

• Processed expression matrices

• Sample metadata

Important GEO Record Types

GEO2R

GEO2R is a web-based tool used to compare groups of samples in a GEO Series.

It can help identify differentially expressed genes.

Important Limitation

A statistically significant expression difference does not independently establish biological function or causation.

The result must be interpreted using:

• Experimental design

• Biological replicates

• Batch effects

• Effect size

• Multiple-testing correction

• Independent validation

17. Expression Atlas

Expression Atlas is an EMBL-EBI resource that provides curated and analysed gene-expression information.

It can be used to investigate:

• Where a gene is normally expressed

• How expression changes under treatment

• Differential expression between conditions

• Tissue-specific expression

• Disease-associated expression

• Expression across organisms

Baseline and Differential Expression

Baseline expression describes gene abundance under normal or reference conditions.

Differential expression describes a change between experimental conditions.

GEO vs Expression Atlas

18. GEO vs SRA

GEO and SRA may contain information from the same sequencing study, but they serve different purposes.

A typical RNA-seq study may use:

• SRA for raw sequencing reads

• GEO for processed expression data

• BioProject for the overall project

• BioSample for biological-sample metadata

III. THREE-DIMENSIONAL STRUCTURE DATABASES

19. Protein Data Bank

The Protein Data Bank or PDB is the global archive of experimentally determined three-dimensional structures of biological macromolecules.

It contains structures of:

• Proteins

• DNA

• RNA

• Protein–nucleic-acid complexes

• Protein–ligand complexes

• Antibodies

• Ribosomes

• Viruses

• Large macromolecular assemblies

The worldwide Protein Data Bank or wwPDB coordinates the global PDB archive.

20. Experimental Methods Represented in PDB

Major structure-determination methods include:

• X-ray crystallography

• Nuclear magnetic resonance spectroscopy

• Cryogenic electron microscopy

• Electron crystallography

• Neutron diffraction

X-Ray Crystallography

X-ray crystallography commonly provides high-resolution atomic models but requires suitable crystals.

NMR Spectroscopy

NMR can determine structures in solution and may provide an ensemble of conformations.

It is generally more practical for smaller molecular systems.

Cryo-Electron Microscopy

Cryo-EM is particularly important for:

• Large macromolecular complexes

• Membrane proteins

• Ribosomes

• Viruses

• Heterogeneous molecular assemblies

21. Information in a PDB Entry

A PDB record may contain:

• PDB identifier

• Macromolecule name

• Organism

• Experimental method

• Resolution

• Atomic coordinates

• Chain identifiers

• Secondary structure

• Ligands

• Metal ions

• Water molecules

• Biological assembly

• Sequence information

• Citation

• Structure-validation information

PDBx/mmCIF is the standard archive format used by wwPDB.

Asymmetric Unit

The asymmetric unit is the smallest portion of a crystal structure from which the complete crystal can be generated using symmetry operations.

Biological Assembly

The biological assembly is the molecular assembly proposed to represent the biologically relevant functional form.

The asymmetric unit and biological assembly are not always identical.

Resolution

Resolution indicates the level of structural detail available from an experimental map.

For comparable crystallographic structures, a lower numerical resolution value generally indicates greater structural detail.

Resolution alone does not guarantee that every region of a structure is equally reliable.

22. Limitations of PDB Structures

A PDB structure should not automatically be treated as a complete representation of the molecule under every biological condition.

Possible limitations include:

• Missing residues

• Missing side chains

• Flexible or disordered regions

• Engineered mutations

• Truncated constructs

• Crystal-packing effects

• Non-physiological ligands

• Uncertain protonation states

• Alternative conformations

• Low local resolution

• Uncertain biological assembly

• Experimental-condition dependence

A structure should be interpreted together with its experimental method and validation report.

23. Experimental PDB vs Predicted Structure Databases

The PDB primarily contains experimentally determined structural models.

Resources such as AlphaFold Protein Structure Database contain computationally predicted structures.

A predicted structure should not be described as an experimentally determined PDB structure.

IV. PATTERN AND SEQUENCE-SIGNATURE DATABASES

24. Sequence Patterns and Signatures

A sequence signature is a characteristic feature used to recognise a protein family, domain or functional site.

Common signature types include:

• Short sequence patterns

• Regular expressions

• Position-specific scoring matrices

• Profiles

• Hidden Markov models

• Fingerprints

• Conserved-domain models

25. Motif, Domain, Family and Site

A short motif may occur by chance.

Therefore, motif detection should be combined with:

• Sequence context

• Domain organisation

• Evolutionary conservation

• Structural information

• Functional evidence

26. PROSITE

PROSITE is a database of protein families, domains and functional sites.

It uses:

• Patterns

• Profiles

• Associated documentation

PROSITE Pattern

A PROSITE pattern describes a conserved sequence using a defined pattern or regular-expression-like representation.

Patterns are useful for detecting short, strongly conserved sequence features.

PROSITE Profile

A profile is a position-specific scoring model that can represent variation across a longer sequence region.

Profiles are generally more sensitive than simple patterns for detecting divergent family members.

ScanProsite

ScanProsite can be used to:

• Scan a protein sequence against PROSITE signatures

• Search a protein database with a selected motif

• Test a user-defined pattern

Limitation of Short Patterns

Short patterns may produce false-positive matches because unrelated proteins may contain the same short sequence by chance.

27. Pfam

Pfam is a database of protein families and domains.

Each Pfam entry is represented using:

• A curated seed alignment

• A multiple sequence alignment

• A profile hidden Markov model

The profile HMM captures:

• Conserved positions

• Variable positions

• Insertions

• Deletions

• Position-specific residue probabilities

Pfam is useful for:

• Detecting conserved domains

• Identifying remote homologues

• Examining domain architecture

• Classifying proteins into families

• Predicting functions of uncharacterised proteins

Pfam Seed and Full Alignments

The seed alignment contains representative, carefully selected sequences used to construct the profile HMM.

The full alignment contains additional sequences detected by the model above the selected inclusion threshold.

Pfam Clans

A Pfam clan groups related Pfam families or domains that may have a common evolutionary origin.

28. InterPro

InterPro is an integrated protein-classification and functional-analysis resource.

It combines predictive signatures from multiple member databases.

InterPro can identify:

• Protein families

• Domains

• Repeats

• Functional sites

• Conserved regions

Important InterPro member resources include:

• Pfam

• PROSITE

• SMART

• PRINTS

• PANTHER

• CDD

• CATH-Gene3D

• SUPERFAMILY

Advantage of InterPro

Different member databases use different detection methods.

By integrating them, InterPro provides a consolidated interpretation of protein-family and domain predictions.

InterPro may also connect protein signatures with:

• Gene Ontology terms

• Structural information

• Functional descriptions

• Pathway information

• UniProt records

29. Other Important Pattern and Domain Databases

PRINTS

PRINTS represents protein families using groups of conserved motifs called fingerprints.

Multiple motifs together are generally more specific than a single short motif.

SMART

SMART focuses on signalling, extracellular and chromatin-associated protein domains and provides domain-architecture analysis.

Conserved Domain Database

The NCBI Conserved Domain Database or CDD contains domain models and supports the detection of conserved domains in protein sequences.

PANTHER

PANTHER classifies genes and proteins according to:

• Families

• Subfamilies

• Molecular functions

• Biological processes

• Pathways

30. PROSITE vs Pfam vs InterPro

V. IMPORTANT DATABASE CONCEPTS

31. Accession Number vs Entry Name

An accession number is a database identifier intended to provide relatively stable access to a record.

An entry name is a human-readable database label and may be less stable.

A sequence version identifies a particular version of the biological sequence.

Important Point

An accession number identifies a database record.

It should not automatically be treated as the biological gene name, protein name or organism name.

32. Database Identifier Mapping

The same biological entity may have different identifiers in different databases.

For example, one protein may have:

• Gene ID

• RefSeq transcript accession

• RefSeq protein accession

• GenBank accession

• UniProt accession

• PDB identifier

• Ensembl gene identifier

Identifier-mapping tools are required to connect records across databases.

Mapping is not always one-to-one because:

• A gene may have several transcripts.

• A transcript may encode different protein isoforms.

• One protein may have multiple structures.

• Database records may represent different sequence versions.

• Databases may merge or split records differently.

33. Common Biological Data Formats

FASTA Format

A FASTA record begins with a header line containing > followed by a sequence identifier and optional description.

The following lines contain the biological sequence.

FASTQ Format

FASTQ contains:

• Sequence identifier

• Nucleotide sequence

• Separator

• Quality-score string

FASTQ should not be confused with FASTA.

FASTA stores sequences, whereas FASTQ also stores per-base quality scores.

34. Database Search Strategy

A suitable database should be selected according to the biological question.

VI. COMMON CSIR-NET TRAPS

1. GenBank is an archival nucleotide-sequence database; RefSeq is a reference collection.

2. GenBank, ENA and DDBJ are partners of INSDC and exchange nucleotide-sequence records.

3. Swiss-Prot contains reviewed records; TrEMBL contains unreviewed records.

4. An unreviewed UniProt record is not automatically an incorrect sequence.

5. UniRef is a sequence-clustering resource, whereas UniParc is a sequence archive.

6. SRA primarily stores raw high-throughput sequencing reads.

7. GEO stores functional-genomics experiments, metadata and processed data.

8. GEO and SRA may represent different data levels from the same experiment.

9. PDB primarily contains experimentally determined structural models.

10. AlphaFold DB contains computationally predicted structures and should not be confused with PDB.

11. The asymmetric unit of a crystal structure is not always the biologically functional assembly.

12. A lower crystallographic resolution value generally represents greater structural detail.

13. A PDB structure may contain missing residues or engineered mutations.

14. PROSITE uses patterns and profiles.

15. Pfam primarily uses profile hidden Markov models.

16. InterPro integrates signatures from multiple protein-family and domain databases.

17. A motif is shorter than a domain and may occur by chance.

18. An accession number is not the same as a gene name.

19. FASTA stores sequences; FASTQ stores sequences plus quality scores.

20. Database annotation should be evaluated using evidence type, review status and biological context.

VII. RAPID REVISION POINTS

• Biological databases store sequences, structures, expression data and functional annotations.

• Primary databases contain submitted or experimental data.

• Secondary databases contain derived patterns, domains or classifications.

• GenBank, ENA and DDBJ form the INSDC collaboration.

• GenBank is an archival nucleotide database.

• RefSeq provides representative reference sequences.

• SRA stores high-throughput sequencing reads.

• UniProtKB is a protein sequence and annotation knowledgebase.

• Swiss-Prot records are reviewed.

• TrEMBL records are unreviewed.

• UniRef reduces sequence redundancy through clustering.

• UniParc archives unique protein sequences.

• GEO stores functional-genomics experiments.

• GPL represents a GEO Platform.

• GSM represents a GEO Sample.

• GSE represents a GEO Series.

• PDB is the global archive of experimentally determined macromolecular structures.

• PDBx/mmCIF is the standard PDB archive format.

• PROSITE uses patterns and profiles.

• Pfam uses multiple sequence alignments and profile HMMs.

• InterPro integrates predictive signatures from multiple databases.

• A motif is a short recurring feature.

• A domain is a larger evolutionary and often independently folding unit.

• An accession number identifies a database record.

• FASTA contains sequences.

• FASTQ contains sequences and quality scores.

VIII. IMPORTANT DISTINCTIONS

IX. FREQUENTLY ASKED QUESTIONS

Which database should be used to obtain reviewed protein annotation?

UniProtKB/Swiss-Prot should generally be preferred when a reviewed protein record is required.

Which database stores raw next-generation sequencing reads?

The Sequence Read Archive stores raw or minimally processed high-throughput sequencing reads.

What is the difference between GenBank and RefSeq?

GenBank is an archival collection of submitted nucleotide sequences. RefSeq provides selected, integrated and relatively non-redundant reference sequences.

Which databases form the INSDC?

GenBank, European Nucleotide Archive and DNA Data Bank of Japan form the International Nucleotide Sequence Database Collaboration.

Which resource is used to retrieve experimental gene-expression studies?

Gene Expression Omnibus is a major repository for functional-genomics and gene-expression experiments.

Which database should be used to retrieve an experimentally determined protein structure?

The Protein Data Bank should be used to retrieve experimentally determined macromolecular structural models.

Which database uses profile hidden Markov models for protein families?

Pfam represents protein families and domains using multiple sequence alignments and profile hidden Markov models.

What is the function of InterPro?

InterPro integrates protein signatures from multiple member databases to classify proteins and predict families, domains, repeats and functional sites.

Can the presence of a sequence motif prove protein function?

No. A motif match may support a functional prediction but should be interpreted with domain context, evolutionary conservation, structure and experimental evidence.

Why are accession numbers important?

Accession numbers provide stable identifiers that allow biological database records to be retrieved, tracked, cited and linked across resources.