BIO/CSC295 2009F, Class 15: Gene Prediction (2)

Admin:
* At least one student who frequents this floor is suspected of having the
  flu.  Please make sure to use the wipes.
* Have a great break!
  * No work for this class during break (other than self review).
  * We'll have a review for the exam Tuesday after break.

Overview:
* Kellis et al. 2003.
* Project 6.5.

Kellis et al. 2003
* Why did Nature give them 14 pages?
  * Most Nature papers are like 3-4 pages (reports 2-3, articles 4-6)
  * They could have separated it into 3 or 4 articles
    * But those would have been so closely tied together that it's awkward
  * Why Nature?
    * A really important method (if we believe it can be used in other
      genomes)
* Why do we study S. cerevisiae? /Why is this a model organism?
  * Not a lot of junk DNA (shotgun works better, easier to study functional
    elements)
  * These four organisms because relatively close
  * Lots of previous research on S. cerevisiae
    * Lets them examine their conclusion in respect to other stuff
    * The SGD
    * Rough estimates of the number of genes studied
  * First Eukaryote sequenced.  
    * But why?
  * Single-cell eukaryote
  * An 'old' organism'
  * It's a very important yeast: Bakers/brewers yeast
    * Bread, beer
    * Yes: It has huge commercial import
    * If you study S. cerevisiae, you can get free beer!
  * Has been used to study a bunch of other things
    * Cell cycle - Dozens of important genes identified
      * Those genes are identical in other organisms
      * Regulation of cell division is important in studying cancer
  * The 'textbook' single-cell eukaryotic organism
  * Whitworth studies intron/exon decisions in it

* What are their primary conclusions?
  * A method for finding likely genes

* Side note:
  * The 'genes' they look at are primarily protein-coding
    * E.g., 'gene and regulatory motif'

* The S. cerevisiae genome was already done.  How did they get the rest?
  * Shotgun: Blast it into bits, sequence the bits, and combine them.
  * How did they blast it into bits?
    * 'Restriction enzyme coctails'
    * Different coctails break it up differently, so that we can 
      get n-fold coverage
  * How do we sequence the bits?  Fun biological techniques
  * How do we assemble?  Shortest superstring problem
    * Often, we use a greedy algorithm: Match the best overlaps first
  * They relied on known assembly algorithms
    * But could also use a known sequence
    * The theory of evolution suggests that the different yeast sequences 
      will be similar
      * And that's a fundamentally important 
* What did they end up after they ran the assembly algorithm?
  * Lots of contigs: Note the whole genome
* [Detour: The divergence between the two least similar yeasts in this
  study is similar to the divergence between human and mice.]
* Exciting: Web lab generated sequences, but the testable hypotheses are
  done completely 'de novo'
  * From descriptive to predictive
  * Not just 'what is' but 'what may be'

ORF analysis
* Large-scale alignment using matches of ORFs (see section titled 
  'genome alignment')
  * We see synteny in the matching regions
    * That is, genes should be in the same alignment in the same chromosome
      * Orientation and order
  * Within the same chunk they should be in the same order
  * Problem: If you're using synteny, are you possibly losing out on
    inversions?
    * Yes, particularly across repeat areas
    * However: Aligning contigs (with 50 or so genes) rather than
      individual genes
    * And they do see some inversions
* See figure 1 for some details of alignment

* Figure 2:
  * Diversification seems to happen near the telomeres rather than
    near the center
  * Telomere is the end of the chromosome, often with repeat sequences
  * Protects the rest of the chromosome
  * There *are* genes near the end of the chromosome, so we can see
    lost genes, repair, and weird things
  * Analysis of telemorase enzyme won the Nobel prize this year
  * How did we get the translocation?
    * tRNAs are similar, making the translocation a bit easier

* What was your favorite discovery in this paper? 
  * Seemed to miss only one gene out of approx. 5K
    * Hand-checked all the named genes to find out that seem to have
      been mistakenly named
    * Seems to be a useful technique
  * Conserved motifs for partners (combinatorial part)
  * A successful and new process
  * Discovery of introns, which normally seems hard to do
  * A 100% conserved gene - it feels like you should have one or two
    differences.
    * Is it evidence of recombination
    * Vida's anecdote: Example of a gene that's 100% conserved across
      species: No observed effect of knockout
* What discoveries do you want to criticize
  * Discovery of regulatory elements
    * Too short for their other technique
    * So they just use some statistics based on the GAL4 sequence
    * Can we really apply it elsewhere?
      * It seems to work ... 
      * And seems like solid theory ...
      * And matches what others found in some cases ...
    * Things need to be conserved to be regulatory

* Hypothesis: Messy genomes are better to analyze, because there's much
  more potential mutation in the garbage, so statistics are easier

* Task:
  * What has been done recently on the yeast MATa2