The Genome Access Course Protein Structure

1
TheGenomeAccessCourseProtein Structure
HSP 70 (1DKG, 1DKZ) and prefoldin (1FXK)
2
Protein structure

• What is the correct amino acid sequence?
• Is the predicted protein complete (ATG real?)
• To be sure - use ORF finder at NCBI

3
ORF finder to BLAST
http//www.ncbi.nlm.nih.gov/gorf/gorf.html
4
Protein Structural Elements

• 2o Structural Elements
• a-Helix
• ß-Sheet
• Globular regions
• Domains
• SH2
• Leucine Zipper

5
Protein function - different categories

• Protein of known function
• Protein of inferred function
• Protein of unknown function

6
Protein of known function

• Work already done
• Ancillary databases (e.g Pubmed. OMIM, MGI, other
  organism specific databases)
• Warning - make sure it really is the SAME protein
• First port of call - LocusLink/Entrez Gene

7
Human genes and OMIM
http//www.ncbi.nlm.nih.gov/entrez/query.fcgi?dbO
MIM
8
Mouse genes and MGI
http//www.informatics.jax.org/
9
And the list goes on.
10
Protein of inferred function

• Similar to protein of known function
• Annotated
• BLAST
• Paralogue (same species) or orthologue (different
  species) or just similar
• Make sure key residues are conserved e.g Pairwise
  or Multiple alignment

11
Protein of inferred function
Human X Chr
12
Protein of Unknown Function

• Not similar at the primary sequence level to a
  protein of known function
• Can you predict function - so many caveats!
• Transmembrane protein?
• TMPred http//www.ch.embnet.org/software/TMPRED_fo
  rm.html
• Protein domains
• can infer function e.g Homeobox
• Warning - some domains are poorly and/or widely
  predicted

13
Domains

• Discrete structural units
• Can infer boundaries from sequence analysis
• 25 500 residues long
• Most lt 200 residues
• Less than 50 residues usually stabilized by SS
  bonds or metal ions

14
LipoxygenaseDomain
gt500 residues
15
WW Domain
33 residues
16
Domain Determination

• Internal duplications
• Detect with a dotplot
• Transmembrane segments
• Hydrophobic, 1535 residues
• Segments easy to predict
• Topology and multiple segments harder to predict
• PHD, TMHMM, TMpred
• Low complexity segments
• Composition typically non-random
• Non-compact folds coiled coils, rods, flexible
  domain linkers
• Complexity function (SEG)
• Small-pitch overlapping repeats (XNU)

17
Protein sequence databases

• Non curated
• Trembl - automatically predicts proteins from CDS
  in Genbank/EMBL/ddBJ
• Entrez protein www.ncbi.nlm.nih.gov80/entrez/quer
  y.fcgi?dbProtein
• Curated
• Swisprot - proteins identified with confidence
  manually added to database
• Uniprot (e.g hosted at EBI http//www.expasy.unip
  rot.org/index.shtml

18
Proteins of Unknown function
Protein domain databases e.g. Interpro
http//www.ebi.ac.uk/interpro/index.html
19
Comparison of Protein Family DBs
Pfam
SMART
CDD
PROSITE
SRS
20

• Conserved Domain Database (NCBI)
• Linked into other NCBI resources
• Includes Pfam and SMART domains (but does not
  necessarily give the same answer)

21
Proteins in Ensembl
22
Proteins in UCSC
23

• HMM family profiles constructed by hand
• Structural data in alignments
• No hierarchy
• No specific compositional bias
• Good graphical output

24
Pfam-A and Pfam-B

• Pfam-A (75)
• Curated, annotated families
• Pfam-B (19)
• Families derived automatically from ProDom
• Other

25

• Protein fingerprint database (fingerprints are
  groups of conserved motifs that characterize a
  protein family)
• Regular grammar for describing profiles (e.g.
  EDQ-x-G-x-DN-A-x-x-GALI)
• Profile search is sensitive, but low coverage
  (signaling)
• Pattern search has high false positive rate

26

• Highly conserved, ungapped MSAs
• Derived from PROSITE

27

• Fingerprints are sets of ungapped weight matrices
• Hierarchical classification for important
  families
• Families, domains, and proteins

28

• Simple Modular Architecture Research Tool
• Collected by Ponting and Bork (641 HMMs)
• Focuses on
• Signaling Domains
• Extracellular domains
• Nuclear domains
• High quality nice graphics

29
Alignment of Representative Members
Profile-HMM built with HMMer 2.0
Search Protein DB
Description
Full alignment
30

• Profiles automatically built from PSI-BLAST
  alignments of Swiss-PROT
• No annotation
• As with other automated DBs (Pfam-B, DOMO),
  useful for seeing if region appears in different
  contexts

31
Protein Sequence Analysis

• Biochemical/biophysical properties
• Secondary Structure
• Super-secondary (signal peptides, domains,
  motifs)
• 3D prediction (Threading)

32
Amphipathic Helix
Edge Strand
Buried Strand
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
Viewing 3D Structures

• Cn3d
• Chime
• RasMol
• Protein Explorer

37
(No Transcript)
38
Protein of inferred or unknown function

• All predictions must be taken as exactly that
• PREDICTIONS!!
• The true function of a protein is NOT known until
  it has been proven in the lab