The Genome Access Course Protein Structure

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

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

3
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

4
LipoxygenaseDomain
gt500 residues
5
WW Domain
33 residues
6
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)

7
Protein Sequence Databases

• GenPept
• Swiss-Prot
• TrEMBL

8
Protein Domain Databases

• Pfam
• PROSITE
• BLOCKS
• PRINTS

• CDD
• ProDom
• SMART
• InterPro

9

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

10
Pfam-A and Pfam-B

• Pfam-A (75)
• Curated, annotated families
• Pfam-B (30)
• Families derived automatically from ProDom
• Other

11

• 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

12

• Highly conserved, ungapped MSAs
• Derived from PROSITE

13

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

14

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

15

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

16
Alignment of Representative Members
Profile-HMM built with HMMer 2.0
Search Protein DB
Description
Full alignment
17

• 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

18

• Pfam, SMART, ProDom, PRINTS, and Prosite domains
• High quality annotation

19
Comparison of Protein Family DBs
Pfam
SMART
CDD
PROSITE
SRS
20
Protein Sequence Analysis

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

21
Amphipathic Helix
Edge Strand
Buried Strand
22
(No Transcript)
23
(No Transcript)
24
Viewing 3D Structures

• Cn3d
• Chime
• RasMol
• Protein Explorer

25
(No Transcript)
26
(No Transcript)
27
Predicting Structure from Sequence

• 100 amino acid protein has 3200 backbone
  configurations
• Threading

28
Protein Structure Prediction is Quite a Challenge
29
3D Structure Prediction

• UCLA-DOE
• SwissMODEL
• CPHmodels

30
Methods for Aligning Structures

• (Double) Dynamic Programming
• Distance Matrix
• Gibbs sampling
• Branch-and-bound searching

31
HMMSTR Local Sequence-Structure Correlations

• Constructed using motif clustering
• No gaps or insertion states
• Non-linear, highly branching model
• Models local motifs common to all proteins