Transcript class11_05

Genes and Genomes
Genome On Line Database (GOLD)
• 243 Published complete genomes
• 536 Prokaryotic ongoing genomes
• 434 Eukaryotic ongoing genomes
December 2004 : 1245 genome projects
Common Genome Browsers
NCBI: http://www.ncbi.nlm.nih.gov/mapview/static/MVstart.html
Eukaryote Only
UCSC: http://genome.ucsc.edu
Ensembl: http://www.ensembl.org
Prokaryote Only
MGV: http://cmbipc49.cmbi.kun.nl/genome/
TIGR: http://www.tigr.org/tigr-scripts/CMR2/CMRHomePage.spl
/
What can we learn from genomes
•
•
•
•
•
•
Genes
Splice variants
Variation analysis
Promoters
Comparative Genomics
Evolution
5’UTR
3’UTR
CDS
Alternative splice variants
Looking for genes in genomes
• Existing mRNA and EST data
• Gene prediction program
• Comparative genomics
ESTs (Expressed Sequence Tags)
• cDNA provide a best tool to identify genes in a genome.
– For unsequenced genomes it was the primary source for
identifying genes
• Basic strategy - select cDNA clones at random and
perform a single automated read from one/both ends of the
transcript.
– Many clones will be redundant.
– Very cost effective.
– ESTs are short (400-600b), relatively inaccurate (2% error).
• ESTs are correlated to known genes using a relatively
small region of sequence alignment.
• Used to discover genes, alternative splicing variants, etc.
Problems with ESTs
-Incomplete Coverage
Bias for high copy number genes
-Experimental mistakes- not always reliable
-Enrichment of 3’ ends of genes
-High representation of cancer cells
Usage of EST
-Predicting of coding regions
-Detecting of alternative splicing
-Clustering to form genes
RefSeq database (NCBI)
• The Reference Sequence (RefSeq) collection aims to
provide a comprehensive, integrated, non-redundant
set of sequences, including genomic DNA, transcript
(RNA), and protein products, for major research
organisms.
• RefSeq standards serve as the basis for medical,
functional, and diversity studies; they provide a
stable reference for gene identification and
characterization, mutation analysis, expression
studies, polymorphism discovery, and comparative
analyses.
Gene Finding Approaches
– Learn characteristics of known genes
– Search for new genes using characteristics
– Different types of genes have different
characteristics.
Prediction Status
The problem of gene prediction is very much
open even in well studied genomes:
The number of genes in human keeps
changing.
Gene Finding
• Input
– Chromosomal genetic sequence
• Output
–
–
–
–
Region which encodes for gene
Strand and reading frame
Start and end of coding sequence
Exon-intron boundaries
Prokaryotes Vs. Eukaryotes
Require different gene finding strategies.
• Prokaryotes:
– the genome is compact (Shorter intergenic regions, no
introns).
– several genes may reside on the same mRNA in
different reading frames.
– Promoter regions are more conserved.
• Eukaryotes
– large genomes; intron/exon structure; alternative
splicing; pseudogenes, very long intergenic regions
– The human genome: average gene ~ 27,800b.
8 exon ~ 100b. intron 100-30,000 b.
ORF Finding
Open Reading Frames –
sequences that presumably code for proteins.
How can ORFs be detected?
• All reading frames are checked.
• Search for initiation and termination codons
within a sequence.
• Are these codons totally conserved?
http://www.ncbi.nlm.nih.gov/gorf/gorf.html
Protein-coding Gene Characteristic
• GC Content
• Uneven codon usage
– Amino acid bias
– Species’ preferred codons
• Promoter and splicing signals
• These characteristics may aid in
– Prediction.
– Validation.
Codon Usage
• DNA is not a random choice of possible
codons for each amino acid.
– It is an ordered list of codons that reflects
evolutionary origin and constraints related to gene
expression.
• Each species has its own coding preferences –
codon usage.
The Genetic Code
• The genetic code - Each amino acid is coded by 3 nucleotides, named codon.
• Code redundancy - Most amino acids are coded by several codons.
- 64 triplets code for 20 amino acids & 3 stop codons.
Codon Usage Database
http://www.kazusa.or.jp/codon/
This site provides:
• Codon usage tables per organism
• Computation of codon usage for query
coding sequences.
Codon Usage Preferences
• Different codon usage for highly vs. weakly expressed
genes.
– in E. Coli genes were divided into 3 groups based on
their codon usage –
- regular genes (70%)
- highly expressed genes (15%)
- horizontally transferred genes (15%)
• There is strong preferences in ORFs for specific codon
pairs and for specific codons near terminators.
• The base in the third position in each codon tends to
repeat itself in the same ORF.
Sequence Signals
• Prokaryotes:
– Promoter (-35, -10 from TSS)
– Ribosome Binding Site (Shine-Dalgarno) is conserved.
Located ~ -15 upstream AUG.
• Eukaryotes
– Transcription signals
TATA (~-30 TSS), cap signal, poly-adenylation site. Any
signal may be missing.
– Translation signals
Kozak signal (immediately upstream ATG).
– Splicing signals – recognized by the Spliceosome. Introns
usually start with GT and end with AG.
Computational Approaches to
Prediction
• Various computational methods including
decision trees, neural nets, Markov models
and Hidden Markov models (HMM).
• A model is studied based on known genes,
and then applied to genomic sequences.
• Each genome defines its own model.
Markov Models
• Probabilistic approach.
• Modeled by states and the probability of
transition from one state to the next.
• The probability of being at state X in step i
depends only on the state we reached at step
i-1.
It has been found that ORFs have a reading-frame specific
hexamer (6mer) composition.
=> the probability of the 6th base can be computed using the
previous 5.
=> The probability that a sequence is an ORF in a specific
reading frame can be computed from its 6th-mer composition.
Grail II for finding Exons
(Neural Network)
Score of 6mers
Score of 6mers in flanking region
Markov model score
GC composition
Exon
score
GC composition in flanking
Score for splicing acceptor
output
Score for splicing donor
Hidden layer
Input layer
GenScan (HMM)
• One of most accurate programs
– Best for human/vertebrate sequences
• Markov parameters for different regions
–
–
–
–
–
Introns beginning at 3 phases
Exons: first, intermediate, last
Promoter region
3’ and 5’ untranslated regions
Intragenic regions
HMM for a GC reach intronic region
Base S1
S2
S3
S4
S5
S6
S7
S8
A
0
0
0.33
0.60
0.49
0.71
0
1
C
0
0
0.37
0.13
0.03
0.07
1
0
G
1
0
0.18
0.14
0.45
0.12
0
0
T
0
1
0.12
0.13
0.03
0.09
0
0
The General Scheme
1. Obtain new genomic DNA sequence.
2. i. Translate in all 6 reading frames and compare
to protein databases.
ii. Perform database similarity search of
expressed sequence tags (EST) database of same
organism, or cDNA sequences if available.
3. Use gene prediction program to locate genes.
4. Analyze regulatory sequences and signals in the
gene. Can help characterize putative genes.
Other gene Finding Tools
• GeneMark (prokaryote, eukaryote)
– http://opal.biology.gatech.edu/GeneMark/
• Glimmer (bacteria, archaea)
– http://www.tigr.org/software/glimmer/
• GeneFinder (human, mouse, arabidopsis)
– http://argon.cshl.org/genefinder/
• HMMgene (vertebrate, C. elegans)
– http://www.cbs.dtu.dk/services/HMMgene/
http://www.tigr.org/genefinding/software.shtml
Prediction Evaluation
Prediction tools are compared using two criteria:
• Sensitivity - % true predicted genes out of the true
genes in the genome.
TP /(TP+FN)
• Specificity - % true predicted non genes out of the
total number of non genes.
TN /(TN+FP)
Both need to be high, results vary from genome to genome
Accuracy comparisons tested on vertebrates
GENSCAN
GRAILII
SN
0.93
0.72
SP
0.93
0.84
Functional RNA Genes
• RNA genes are transcribed but are not translated – no
codon preference exists.
How can rRNA, tRNA and small RNA genes be
predicted?
• Promoter regions can be characterized, but remain a
big challenge.
• RNA secondary structure is important.
Can be predicted using RNA structure
prediction tools (MFOLD tool).
Comparative genomics
• Finding Orthologs
• Looking for genes in one species not found
in another
• Searching for conserved regulatory
elements
• Gene Clusters
• Conserved regulatory networks
Conservation of the IGFALS (Insulin-like growth factor)
Between human and mouse.