1. Sequence Based Analysis Tutorial
NIH Proteomics Workshop
Cecilia Arighi, Ph.D.
Protein Information Resource at
Georgetown University Medical Center

2. Retrieval, Sequence Search & Classification Methods
Retrieve protein info by text / UID
Sequence Similarity Search
BLAST, FASTA, Dynamic Programming
Family Classification
Patterns, Profiles, Hidden Markov Models, Sequence Alignments, Neural Networks
Integrated Search and Classification System

3. Sequence Similarity Search (I)
Based on Pair-Wise Comparisons
Dynamic Programming Algorithms
Global Similarity: Needleman-Wunch
Local Similarity: Smith-Waterman
Heuristic Algorithms
FASTA: Based on K-Tuples (2-Amino Acid)
BLAST: Triples of Conserved Amino Acids
Gapped-BLAST: Allow Gaps in Segment Pairs
PHI-BLAST: Pattern-Hit Initiated Search
PSI-BLAST: Position-Specific Iterated Search

4. Sequence Similarity Search (II)
Similarity Search Parameters
Scoring Matrices – Based on Conserved Amino Acid Substitution
Dayhoff Mutation Matrix, e.g., PAM250 (~20% Identity)
Henikoff Matrix from Ungapped Alignments, e.g., BLOSUM 62
Gap Penalty
Search Time Comparisons
Smith-Waterman: 10 Min
FASTA: 2 Min
BLAST: 20 Sec 10

5. Feature Representation
Features of Amino Acids: Physicochemical Properties, Context (Local & Global) Features, Evolutionary Features
Alternative Amino Acids: Classification of Amino Acids To Capture Different Features of Amino Acid Residues

6. Substitution Matrix
Likelihood of One Amino Acid Mutated into Another Over Evolutionary Time
Negative Score: Unlikely to Happen (e.g., Gly/Trp, -7)
Positive Score: Conservative Substitution (e.g., Lys/Arg, +3)
High Score for Identical Matches: Rare Amino Acids (e.g., Trp, Cys) 10

7. Secondary Structure Features
a Helix Patterns of Hydrophobic Residue Conservation Showing I, I+3, I+4, I+7 Pattern Are Highly Indicative of an a Helix (Amphipathic)
b Strands That Are Half Buried in the Protein Core Will Tend to Have Hydrophobic Residues at Positions I, I+2, I+4, I+6

8. BLAST BLAST (Basic Local Alignment Search Tool) Extremely fast Robust
Most frequently used
It finds very short segment pairs (“seeds”) between the query and the database sequence
These seeds are then extended in both directions until the maximum possible score for extensions of this particular seed is reached

9. BLAST Search From BLAST Search Interface
Table-Format Result with BLAST Output and SSEARCH (Smith-Waterman) Pair-Wise Alignment
Links to iProClass and
UniProtKB reports
Link to NCBI
taxonomy
Link to PIRSF
report
Click to see
SSearch alignment
Click to see
alignment

10. Blast Result & Pairwise Alignment
BLAST Aligment

11. Classification What is classification?
Why do we need protein classification?
Different levels of classification
Basis for functional protein classification
How to classify a protein of unknown function?

12. Classification Databases
C - x(2,4) - C - x(3) - [LIVMFYWC] - x(8) - H - x(3,5) - H The 2 C's and the 2 H's are zinc ligands
Protein motif
Protein domain
3-D structure
Whole-protein
Group proteins according to the presence of a common domain
Group proteins according to common 3D structure
Group proteins according to common domain architecture and length
Protein Domain: Structurally compact, independently folded unit that forms a stable 3D-structure and shows a certain level of evolutionary conservation
Protein motif: A set of conserved amino acid residues that are important for protein function and located within a certain distance from one another

13. Family Classification Methods
Based on Other Classification Information
Multiple Sequence Alignment (ClustalW)
ProSite Pattern Search
Profile Search
Hidden Markov Models (HMMs)
Domain (Pfam); Whole protein (PIRSF)
Neural Networks

14. How do you build a tree? Pick sequences to align Align them
Verify the alignment
Keep the parts that are aligned correctly
Build and evaluate a phylogenetic tree
Integrated Analysis

15. Multiple Sequence Alignment: CLUSTALW
Pairwise alignment:
Calculate distance matrix
Mean number of differences per residue
Unrooted Neighbor-Joining Tree
Branch length drawn to scale
Rooted NJ Tree (guide tree)
Root place at a position where the means of the branch lengths on either side of the root are equal
Progressive Alignment guided by the tree
Alignment starts from the tips of the tree towards the root
Thompson et al., NAR 22, 4675 (1994).

16. PIR Multiple Alignment and Tree
From Text/Sequence Search Result or CLUSTAL W Alignment Interface

17. Here is an example of two different functions easily separated on a phylogenetic tree. Each functional group is used to build an HMM.

18. PIR Pattern Search Signature Patterns for Functional Motifs
From Text/Sequence Search Result or Pattern Search Interface
P-[IV]-[WY]-x(3)-H-[MR]-V-x(3,4)-Q-x(1,2)-D-x(4,5)-G-A-N
Alignment of a region involved
in catalytic activity
Create Pattern and search in database: A
P-[IV]-[WY]-x(3)-H-[MR]-V-x(3,4)-Q-x(1,2)-D-x(4,5)-G-A-N B
Test sequence against PROSITE database
O05689

19. Pattern Search Result (I)
One Query Pattern Against UniProtKB or UniRef100 DBs
Display the query
pattern
Indicate pattern sequence region(s)
Links to iProClass and
UniProtKB reports
Link to NCBI
taxonomy
Link to PIRSF
report

20. Pattern Search Result (II)
One Query Sequence Against PROSITE Pattern Database

21. Profile Method
Profile: A Table of Scores to Express Family Consensus Derived from Multiple Sequence Alignments
Num of Rows = Num of Aligned Positions
Each row contains a score for the alignment with each possible residue.
Profile Searching
Summation of Scores for Each Amino Acid Residue along Query Sequence
Higher Match Values at Conserved Positions

22. Prosite PS50157 profile for Zinc finger C2H2

23. 1
PIRSF scan
Shows PIRSF that the query belongs to
Search One Query Protein Against all the Full-length and Domain HMM models for the fully curated PIRSFs by HMMER
The matched regions and statistics will be displayed.
Statistical data for all domains
Statistical data per domain
Alignment with consensus sequence

24. Creation and Curation of PIRSFs

25. Integrated Bioinformatics System for Function and Pathway Discovery
Data Integration
Associative Analysis

26. Analytical Pipeline Family Classification & Functional Analysis
Query Sequence
UniProt
Top-Matched Superfamilies/Domains
BLAST Search
HMM Domain Search
Predicated Superfamilies/Domains/Motifs/Sites/SignalPeptides/TMHs
SSEARCH
CLUSTALW
Superfamily/Domain/Motif Alignments
Family Relationships & Functional Features
Family Classification & Functional Analysis
HMM Motif Search
Pattern Search
SignalP/TMHMM
Analytical Pipeline

27. Integrated Bioinformatics System
Global Bioinformatics Analysis of 1000’s of Genes and Proteins
Pathway Discovery, Target Identification

28. Lab Section

29. Rat eye lens phosphoproteomics in normal and cataract
Kamei et al., Biol. Pharm. Bull., 2005.
Normal
Cataract
(-) pI (+)
More phosphorylated spots in cataract sample.
Digestion and MS from Spot 16 gave these peptides:
MDVTIQHPWFKR
ALGPFYPSR
CSLSADGMLTFSG
YRLPSNVDQSALS Mw
MDVTIQHPWFKR
We want to identify the protein(s) that contain these peptides
Use Peptide Search

30. Peptide Search
Restrict search to
an organism

31. Peptide Search & Results
Species restricted search
Sorting arrows
Search in UniProtKB, 23 proteins
Links to iProClass and
UniProtKB reports
Link to NCBI
taxonomy
Link to PIRSF
report
Matching peptide
highlighted in
the sequence

32. Retrieve more sequences
Batch Retrieval Results (I)
Retrieve multiple proteins in from iProClass using a specific identifier or a combination of them
Provides a means to easily retrieve and analyze proteins when the identifiers come from different databases
Retrieve more sequences

33. Blast Similarity Search
What proteins are related to rat CRYAA?
Perform sequence similarity search
>P24623

34. Blast Search Results
BLAST (partial) result for CRYAA_RAT in UniProtKB database
BLAST alignment with the human protein

35. Pairwise Alignment

36. PIR protein family classification
PIR Text Search (
UniProtKBDatabase
and unique UniParc
sequences
Let’s search for human crystallins
PIR protein family classification
database

37. Let’s look for crystallins which have 3D structure
Refine your search or start over
Display PDB ID

38. Let’s perform a multiple alignment on the sequences containing PF00030
Domain Display allows to compare simultaneously Pfam domains present in multiple proteins
Share same domain
architecture
Let’s perform a multiple alignment on the sequences containing PF00030

39. Multiple Alignment

40. Interactive Phylogenetic Tree and Alignment
Beta B1 and gamma crystallins share the same domains, SCOP fold and share significant sequence similarity suggesting that they are related

41. Pattern Search (I) Select P07320 and perform a pattern search
Search for proteins containing this pattern (PS00225) in rat

42. Pattern Search Result
Beta and gamma Crystallins have multiple copies of this pattern

43. Pfam domains assigned with high confidence
PIRSF provides a single platform where all the previous analysis has been done by curators
Pfam domains assigned with high confidence
Validation tag
Represents extent of manual curation
Link to
PIRSF report

44. Taxonomic
Distribution
Alpha-crystallin is exclusively found in metazoans
Domain
Architecture
Multiple Alignment

45. PIRSF scan

46. PIRSF report (I): a single platform to study proteins
Subfamily level

47. PIRSF report (II) http://www.geneontology.org/
Cross-links to other databases
provides a controlled vocabulary to describe gene and gene product attributes in any organism

48. alpha-Crystallin and Related Proteins
Alpha crystallin beta chain
HSPs
Alpha crystallin alpha chain