Part I: Identifying sequences with

1
Part IIdentifying sequences with
Speaker S. Gaj
Date 11-01-2005
2
Annotation

• Annotation
• Best possible description available for a given
  sequence at the current time.
• How to annotate?
• Combining
• Alignment Tools
• Databases
• Datamining (scripts)

Background
3
Microarrays
4
(No Transcript)
5
(No Transcript)
6
Introduction

• Global alignment
• Optimal alignment between two sequences
  containing as much characters of the query as
  possible.
• Ex predicting evolutionary relationship
  between genes,
• Local alignment
• Optimal alignment between two sequences
  identifying identical area(s)
• Ex Identifying key molecular structures
  (S-bonds, a- helices, )

Background
7
Introduction

• Basic Local Alignment Search Tool
• Aligning an unknown sequence (query) against all
  sequences present in a chosen database based on a
  score-value.
• Aim
• Obtaining structural or functional information
  on the unknown sequence.

BLAST
8
Programs

• Different BLAST programs available
• Usable criteria
• E-Value, Gap Opening Penalty (GOP), Gap Extension
  Penalty (GEP),
• Terms
• Query Sequence which will be aligned
• Subject Sequence present in database
• Hit Alignment result.

BLAST
9
Common BLAST problems

• BlastN

Clone seq
mRNA
Sequencing Error
BLAST

• Solution
• Low penalty for GOP and GEP 1

10
Translation Problems

• 6-Frame translation

gtemblJ03801HSLSZ Human lysozyme mRNA, complete
cds with an Alu repeat in the 3' flank.
BLAST
L A L P S S Q H
E G S H C S G A
1
ctagcactctgacctagcagtcaacatgaaggctctcattgttctggggc
t...
11
Translation Problems

• 6-Frame translation

gtemblJ03801HSLSZ Human lysozyme mRNA, complete
cds with an Alu repeat in the 3' flank.
3
2
H S D L A V N M
K A L I V L G
BLAST
L A L P S S Q H
E G S H C S G A
1
ctagcactctgacctagcagtcaacatgaaggctctcattgttctggggc
t...
-1
-2
-3
12
Common BLAST problems
intron
exon
Gene X
Translation
BLAST
full mRNA
Splicing
mRNA
13
Common BLAST problems
mRNA
Clones derived from mRNA
BLAST
BlastX against protein sequence
3 possible hit-situations
14
Common BLAST problems

• ? Yields no protein hit

? Aligns with protein in 1 of the 6 frames.
BLAST
? Part perfect alignment
or
15
Part II Databases and annotation
16
Introduction

• Primary database
• DNA Sequence (EMBL, GenBank, )
• AminoAcid Sequence (SwissProt, PIR, )
• Protein Structure (PDB, )
• Secondary database
• Derived from primary DB
• DNA Sequence (UniGene, RefSeq, )
• Combination of all (LocusLink, ENSEMBL, )
• Structure
• Flat file databases

Databases
17
Primary Databases

• EMBL
• DNA Sequence
• Human 4.126.190.851 nucleotides in 292.205
  entries
• Clones, mRNA, (Riken) cDNA,
• New sequences can be admitted by everyone.
• No curative check before admittance.

Databases
18
Primary Databases

• SwissProt
• Amino Acid sequence
• Human
• Contains protein information
• SwissProt ?(EU) ? PIR (USA)
• Crosslinks to most informative DB (PDB, OMIM)
• Part of UniProt consortium.
• Each addition needs validation by appointed
  curators.
• Highly curated

Databases
19
Secondary Databases

• TrEMBL
• Translated EMBL
• Hypothetical proteins
• After careful assessment ? SpTrEMBL ? SwissProt

Databases
20
Secondary Databases

• UniGene
• Automated clustering of sequences with high
  similarity
• Derived from GenBank / EMBL
• 1 consensus-sequence
• Species-specific

Databases
21
Secondary Databases

• LocusLink
• Curated sequences
• Descriptive information about genetic loci
• RefSeq
• Non-redundant set of sequences.
• Genomic DNA, mRNA, Protein
• Stable reference for gene identification and
  characterization.
• High curation

Databases
22
Database Quality?
mRNA
Protein
DNA
EMBL
SwissProt
Databases
Submitter
Submitter
Curators
Database Manager
Database Manager
23
How to Annotate?

• BlastN against random nucleotide DB
• ESTs
• BlastN against structured nucleotide DB
  (UniGene,
  RefSeq)
• mRNA hits
• Sometimes not annotated at all
• Best information

Databases
24
Microarrays
25
(No Transcript)
26
(No Transcript)
27
Part III Annotation Techniques
28
What do we have?

• Probe sequence
• Alignment Tools (e.g. BLAST)
• Databases
• !?! What to choose ?!?

Annotation
29
Possibilities?

• 1. Do it like everyone else does.
• 2. Make use of curative properties of certain
  databases
• Goal
• Annotate as many genes with as much information
  as possible (e.g. SwissProt ID)

Annotation
30
1st Approach - General

• Done by most array manufacturers
• Step-by-step approach
• BLAST sequences against nucleic database
  
  (preferably UniGene)
• Extract high quality (HQ) hits (gt95)
• For each HQ hit search crosslinks.
• Find a well-described (SwissProt) ID for each
  sequence.

Annotation Techniques
31
1st Approach - Concept
Annotation Techniques
32
2nd Approach - General

• Make use of present database curation
• Other way around
• Use SwissProt to clean out EMBL
• Result
• Cleaned EMBL database with direct SP
  crosslinks
• BLAST against cEMBL
• Extract high quality alignment hits (gt95)
• Convert EMBL ID to SP ID.

Annotation Techniques
33
2nd Approach - Concept
Annotation Techniques
34
Annotating Incyte Reporters

• Total 13.497
• cEMBL-approach 2.898 (21,47) SP-IDs
• DM approach 10.013 (74,18) UG-IDs in which
• M 4.723 (34,9) SP-IDs MR 5.147 (38,1)
  SP-IDs MRH 6.641 (49,2) SP-IDs

Results
35
Annotating Incyte Reporters

• All reporters present on Incyte Mouse UniGene 1
  converted
• Total 9.596 reporters
• Old annotation 9.370 (97,6) UG-IDs in which
• Non-existing UG-IDs 5.713 (59,5) M 1.939
  (20,2) SP-IDs
• MR 2.096 (21,8) SP-IDs MRH 2.582 (26,9)
  SP-IDs
• Datamining approach 8.532 (88,9) UG-IDs in
  which
• M 4.145 (43,2) SP-IDs MR 4.499 (38,1)
  SP-IDs MRH 5.576 (60,1) SP-IDs
• Custom EMBL-approach 2.898 (30,2) SP-IDs

Results
36
Annotating Incyte Reporters

• Combined methods Incyte Mouse UniGene 1
  reporters
• Total 9.596 reporters
• No annotation 1.062 (11) reporters
• Annotated with SP-ID 5.895 (61,3) reporters of
  which
• 2.184 (22,7) identical SP-IDs 532 (5)
  reporters with improved SP-IDs by EMBL-method
• 174 (1,8) reporters with different mouse SP-IDs
  5 reporters found only by EMBL-method

Results
37
Conclusions

• Annotation is much needed
• Array sequences can point to different genes
• Direct translation into protein not best option
• Sequencing errors
• Addition or deletion of nucleotides
• 6-Frame window
• Public nucleotide databases are redundant.
• Sequencing errors
• Differences in sequence-length
• Attachment of vector-sequence

Conclusions
38
Questions?
End