Lecture 6: Gene Prediction

1
Lecture 6 Gene Prediction

• Chapter 6
• Eukaryotic gene and intron splicing prediction
• Genome and gene organization

2
Eukaryote Gene Prediction

• Its like finding a 2gram needle in a 6,000kg
  haystack.

3
Finding Gene-Rich Regions in whole genome

• Isochores (high density versus low density)
• CpG islands
• Codon biased regions
• Mask the junk (mask the repeats)

4
Isochores

• Genome organization
• Nucleotides?genes?isochores?chromosomes
• Just another level of organization to help
  organize our information!
• Isochore
• 1 mb
• Homogenous bp composition (GC content uniform in
  individual isochores)
• Human example
• 5 different classes of isochores
• L1, L2, H1, H2, H3
• Llow density (lower GC and lower gene content)
• Hhigh density (higher GC and higher gene
  content)

Where would you Start to sequence A genome?
5
CpG Islands

• Randomness of most dinucleotide pairs
• CpG found at 20 of the normal random frequency
  overall
• Some regions are more dense with CpG than others
• CpG Islands
• Higher density of CpG in a given area as compared
  to overall percentage of CpG
• CpG islands occur at specific sites
• 45,000 islands in human genome
• Found typically in housekeeping genes and at many
  regulatory binding sites or promoter regions
• 1-2kb at 5 ends of many human genes
• About 4,500 islands in human genes
• Never found in introns or in junk DNA or in gene
  free regions

6
Codon Bias

• Every organisms seems to prefer to use some
  codons over others within genes
• Yeast AGA for arginine 48 of time
• Codon bias occurs in exons, but not in introns
• This can be used as more evidence for the
  presence of an intron

7
Codon Usage in a Properly Spliced Gene
8
Locating Genes and Coding Regions in Eukaryote
Gene-Rich Regions

• ORF locating versus Coding Region Location
• Note Easy!
• Large genomes
• About 1 of human DNA encodes functional genes.
• Alternatively, 85 of bacteria DNA encodes
  functional genes
• Genes are interspersed among long stretches of
  non-coding DNA.
• More than one chromosome and more than one copy
  of a chromosome
• Repeats, pseudo-genes, and introns confound
  matters
• Will take at least 15 years to find all genes in
  a given genome once genome is sequenced.

9
Eukaryote Prediction

• A variety of features must be found and the
  features must be found in specific locations
• Any one feature (ie finding a promoter region)
  cannot be used by itself to predict a functional
  gene
• Prediction algorithms have an accuracy of about
  50 for Euks
• Focus for finding genes
• Promoter elements
• Regulatory binding sites
• Focus for finding coding regions
• Splice Junctions
• Codon bias
• All DNA motiffs observed in 5 to 3 direction on
  coding/sense strand

10
Finding Genes Promoter Elements

• Binding of transcriptional factors to promoter
• Assist in RNA polymerase II binding
• TATA box
• TATAWAW, where WA or T
• At 25 of trc start site
• Inr sequence (initiator sequence)
• YYCARR, where YC or T and RG or A
• At 1 from TATA box
• Subtle differences in these basal trc factors
  make them unreliable by themselves
• Need more evidence

11
Finding Genes Regulatory Protein Binding Sites

• Found upstream from trc start site
• Regulatory proteins bind to DNA and promote or
  prevent trc
• CAAT box found 80 from actual gene
• Enhancers
• EX) GGGCGG site ? binding site for Sp1 protein
• -500 to 500 to trc start site
• Bend DNA?change DNA orientation to initiate trc

12
Finding Coding Regions Within Genes Splice
Junctions

• Junctions between intron and 2 flanking exons
• GT-AG rule (usually)
• First 2 nucleotides of intron in mRNA is GU
  (coding DNA GT)
• Last 2 nucleotides of intron is AG
• Introns must be at least 60 bp, but there are no
  limits on length
• No limits on distribution
• Other splice junctions exist and differ for
  different genes
• Alternative splicing complicates things here.
• 20 of all human genes!

13
Finding Coding Regions Within Genes Splice
Junction
Intron
G100 T100 A62 A68 G64T6312-C/T N C68 A100
G100
Exon
Exon
3 Splice Junction (5-AG-3)
5 Splice Junction (5-GT-3)

• Subscript refers to percentage of times you see
  that particular base.
• Dots represent length of intron (can be any
  number of nucleotides greater than about 60).
• Consensus GTAAGT.YYYYYYYYYYYYNCAG

14
Finding Coding Regions Within Genes Alternative
Splicing
Mouse Troponin T Cardiac muscle or skeletal
muscle
Intron
Intron
Tnni3 Cardiac
Intron
Intron
Tnnt1 Skeletal
15
Finding Coding Regions Within Genes Coding Bias

• Also used for predicting Coding regions

16
Gene Prediction SoftwareGreat for Eukaryotes

• Genescan and HMM Gene
• Used to predict exon locations and repeated
  elements
• Splices exons and translates (when more than one
  present) so you can do a BLASTP
• http//genes.mit.edu/GENSCAN.html
• http//www.cbs.dtu.dk/services/HMMgene/
• Used primarily for Human/vertebrate genomic
  sequences
• Not good for invertebrate sequences
• Practice Together