Transcript PSI - Bioinformatics Training Network (BTN)

UniProtKB
Sandra Orchard
EBI is an Outstation of the European Molecular Biology Laboratory.
Importance of reference protein sequence
databases
• Completeness and minimal redundancy
A non redundant protein sequence database, with maximal coverage
including splice isoforms, disease variant and PTMs.
Low degree of redundancy for facilitating peptide assignments
• Stability and consistency Stable identifiers and consistent nomenclature
Databases are in constant change due to a substantial amount of work to
improve their completeness and the quality of sequence annotation
• High quality protein annotation
Detailed information on protein function, biological processes, molecular
interactions and pathways cross-referenced to external source
Summary of protein sequence databases
Database
Description
Species
UniProtKB
Expertly curated section (UniProtKB/Swiss-Prot) and computerannotated section (UniProtKB/TrEMBL); minimum level of redundancy;
high level of integration with other databases; stable identifiers;
diversity of sources including large scale genomics, small scale
cloning and sequencing, protein sequencing, PDB, predicted
sequences from Ensembl and RefSeq
Many
UniRef100
Assembled from UniProtKB, Ensembl and RefSeq; merges 100%
identical sequences; stable identifiers
Many
Ensembl
Predictions using automated genome annotation pipeline; explicitly
linked to nucleotide and protein sequences; stable reference; merge
their annotations with Vega annotations at transcript level; extensive
quality checks to remove erroneous gene models ; high level of
integration with other databases
Over 50 Eukaryotic genomes
Expanded Ensembl Metazoa,
Plants and Fungi, Protists,
Bacteria and Archaea
RefSeq
NCBI creates from existing data; ongoing curation; non-redundant;
explicitly linked nucleotide and protein sequences; stable reference;
high level of integration with other databases
Limited to fully sequenced
organisms
IPI
Good balance between degree of redundancy and completeness;
references to primary data sources; attempts to maintain stable
identifiers; Assembled from UniProtKB, RefSeq and Ensembl
7 eukaryotic genomes
Entrez protein (NCBInr)
Assembled from GenBank and RefSeq coding sequence translations
and UniProt KB ; annotations extracted from source curated
databases; high degree of sequence redundancy
Many
Updated from Nesvizhskii, A. I., and Aebersold, R. (2005) Interpretation of shotgun proteomic data: the protein inference problem. Mol. Cell. Proteomics. 4,1419–1440l
UniProtKB

UniProt Knowledgebase:

1.
2.
2 sections
UniProtKB/Swiss-Prot Non-redundant, highquality manual annotation - reviewed
UniProtKB/TrEMBL Redundant, automatically
annotated - unreviewed
www.uniprot.org
Master headline
Collaboration projects between database
resources
• CCDS project, a collaboration between Ensembl, NCBI,
UCSC and UniProt, aims to provide a standard set of
gene predictions for the human and mouse genomes
• Considerable communication effort between curators from
different groups is on-going
• Ensembl and UniProt collaboration to cover the gaps in
gene predictions in UniProtKB (one sequence for each
protein coding transcript in Ensembl)
• Ensembl high quality gene/transcript models (quality checks
remove gene models with erroneous structures or supported by
dubious evidence – e.g. cDNA fragments with short/wrongly
annotated ORF)
• UniProtKB high quality protein sequences
Complete proteome data sets in UniProtKB
• Ensembl sequences have now been incorporated for an
increasing number of species: human, mouse, rat ,
zebrafish, chicken, dog pig and cow.
• Tagged with Complete proteome keyword by release
2011_06 of 31st May
• Fasta files by FTP
• One file per species containing canonical + isoform sequences
Manual annotation of the human proteome
(UniProtKB/Swiss-Prot)
• A draft of the complete human proteome has been
available in UniProtKB/Swiss-Prot since 2008
• Manually annotated representation of 20,252 protein
coding genes with over 35,000 protein sequences - an
additional 41,000 UniProtKB/TrEMBL form the complete
proteome set
• Approximately 63,000 single amino acid polymorphisms
(SAPs), mostly disease-linked
• 80,000 post-translational modifications (PTMs)
• Close collaboration with NCBI, Ensembl, Sanger Institute
and UCSC to provide the authoritative set to the user
community
Manual annotation of UniProtKB/Swiss-Prot
Splice variants
Sequence
Sequence
features
UniProtKB
Ontologies
Annotations
Nomenclature
References
Sequence curation, stable identifiers, versioning and archiving
For example – erroneous gene model predictions, frameshifts
….
..premature stop codons, read-throughs, erroneous initiator methionines…..
Master headline
Splice variants
Master headline
Identification of amino acid variants
..and of PTMs
… and also
Master headline
Domain annotation
Binding sites
Master headline
Protein nomenclature
Master headline
Master headline
Annotation - >30 defined fields
Controlled vocabularies used whenever possible…
Master headline
..and also imported from external resources
Binary interactions taken from the IntAct database
Interactors of human p53
Master headline
Controlled vocabulary usage increasing – for example
from the Gene Ontology
Annotation for human Rhodopsin
Master headline
Sequence evidence
Type of evidence that supports the existence of a protein
1 Evidence at protein level
There is experimental evidence of the existence of a protein
(e.g. Edman sequencing, MS, X-ray/NMR structure, good quality
protein-protein interaction , detection by antibodies)
2 Evidence at transcript level
The existence of a protein has not been proven but there is
expression data (e.g. existence of cDNAs, RT-PCR or Northern blots)
that indicates the existence of a transcript.
3 Inferred from homology
The existence of a protein is likely because orthologs exist in closely
related species
4 Predicted
5 Uncertain
UniProtKB/TrEMBL
Multiple entries for the same protein (redundancy)
can arise in UniProtKB/TrEMBL due to:
o Erroneous gene model predictions
o Sequence errors (Frame shifts)
o Polymorphisms
o Alternative start sites
o Isoforms
Apart from 100% identical sequences all merged
sequences are analysed by a curator so they can be
annotated accordingly.
Automatic Annotation
•Automated clean-up of annotation from original
nucleotide sequence entry
• Additional value added by using automatic
annotation
•Recognises common annotation belonging to a
closely related family within UniProtKB/Swiss-Prot
• Identifies all members of this family using
pattern/motif/HMMs in InterPro
• Transfers common annotation to related family
members in TrEMBL
Master headline
InterPro: the ingredients...
CL0247
CL0181
...etc
Users can search using
e.g. sequence, UniProt
identifier, PDB id...
Protein Sequence Characterisation
BLAST
Basic information
more sequences
Build up consensus sequences of
families, domains, motifs or sites
Conserved signatures
Patterns
Pattern/motif in sequence  regular expression
Can define important sites
EXAMPLE: PS00262 Insulin family signature
B chain xxxxxxCxxxxxxxxxxxxCxxxxxxxxx
A chain xxxxxCCxxxCxxxxxxxxCx
|
|
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLV
EALYLVCGERGFFYTPKTRREAEDLQVGQVELGGGPG
AGSLQPLALEGSLQKRGIVEQ CCTSICSLYQLENYC N
Patterns
Pattern/motif in sequence  regular expression
Can define important sites
EXAMPLE: PS00262 Insulin family signature
B chain xxxxxxCxxxxxxxxxxxxCxxxxxxxxx
A chain xxxxxCCxxxCxxxxxxxxCx
|
|
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLV
EALYLVCGERGFFYTPKTRREAEDLQVGQVELGGGPG
AGSLQPLALEGSLQKRGIVEQ CCTSICSLYQLENYC N
Regular expression
C-C-{P}-x(2)-C-[STDNEKPI]-x(3)-[LIVMFS]-x(3)-C
Fingerprints
Several motifs  characterise family
Identify small conserved regions in divergent proteins
Different combinations of motifs describe subfamilies
EXAMPLE: PR00107 Phosphocarrier HPr signature
PTHP_ENTFA:
MEKKEFHIVAET GIHARPATLLVQTASK FNSDINLEY KGKSVNLK
SIMGVMSL GVGQGSDVTITVDGADE AEGMAAIVETLQKEGLAE
His phosphorylation site
Conserved site
Ser phosphorylation site
Fingerprints
Sequence alignment
Define motifs
Extract motif
sequences
Fingerprint
signature
PR00000
His
phosphorylation
site
xxxxxx
xxxxxx
xxxxxx
xxxxxx
Ser
Conserved
phosphorylation site
site
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
Correct order
1
2
3
Correct spacing
Hidden Markov Models (HMM)
Can characterise protein over entire length
Models conserved and divergent regions
(position-specific scoring)
Models insertions and deletions
 Outperform in sensitivity and specificity
 More flexible (can use partial alignments)
Hidden Markov Models (HMM)
Sequence
alignment
Sequence 1:
Sequence 2:
Sequence 3:
Sequence 4:
Sequence 5:
Sequence 6:
Sequence 7:
Scoring
matrix
(residue frequency at
each position in
alignment)
Bayesian statistics
probability scoring
Profile
Hidden Markov Models (HMM)
HMMER package (http://hmmer.janelia.org/)
HMMbuild
HMMcalibrate
Database search
SAM Profile HMMs
Homologous structural superfamilies
Start with single seed sequence
Create 1 model for
every protein in
superfamily 
combine results
Master headline
Few proteins in family
have PDB structures
Proteins in superfamily
may have low
sequence identity
InterPro Entry
Groups similar
signatures together
Links related
signatures
Adds extensive annotation
Linked to other databases
Structural information and viewers
What we add at InterPro
Quality control
32
18.07.2015
The InterPro entry types
Family
Proteins share a common evolutionary origin, as reflected in their
related functions, sequences or structure
Domain
Distinct functional, structural or sequence units that may exist in a
variety of biological contexts
Repeats
Short sequences typically repeated within a protein
Sites
PTM
33
18.07.2015
Active
Site
Binding
Site
Conserved
Site
Master headline
Searching UniProt – Simple Search
• Text-based searching
• Logical operators ‘&’ (and), ‘|’
Master headline
Searching UniProt – Advanced Search
Master headline
Searching UniProt – Search Results
Each linked to the UniProt entry
Master headline
Searching UniProt – Search Results
Master headline
Searching UniProt – Search Results
Master headline
Searching UniProt – Blast Search
Master headline
Searching UniProt – Blast Search
Master headline
Searching UniProt – Blast Results
Alignment with query sequence
Master headline
Searching UniProt – Blast Results
Master headline
Finding a complete proteome in UniProtKB
Complete
Proteomes