This is an interesting and controversial topic. The unexpectedly low number of genes in the human genome (~32000, which is roughly twice the number for Drosophila and worm) raised again the long-standing question in biology:
what is the source for organism complexity?
After it was discovered that alternative splicing, a mechanism for increasing transcript diversity, is actually widespread in the human genome, it appeared reasonable to think that alternative splicing might be one of the major contributors for human genome complexity.
Surprisingly, work from Peer Bork lab argued against this hypothesis. They found that after correcting for EST coverage, the frequency of alternative splicing seemed to be comparable among animals , and humans didn't appear to have a substantially higher rate of alternative splicing:
Alternative splicing and genome complexity. Brett D, Pospisil H, Valcarcel J, Reich J, Bork P. Nat Genet. 2002 Jan;30(1):29-30.
There is certain amount of doubt concerning this interesting and apparently counter-intuitive conclusion. In a recent correspondence published on Nature Genetics Kim H et.al. used an interesting but indirect method to re-evaluate this question. By estimating the number of total transcripts in different genomes, they suggested that human actually has more alternative splicing than mouse and fruitfly.
However, although Kim H et.al. claimed that their method is independent on EST coverage, in their reply Bork P and colleagues showed that the method used by Kim H et.al. was still dependent on the amount of EST coverage. For example, their method showed that there is substantially lower rate of alternative splicing in rat than mouse.
I believe there will still be more debates regarding this interesting question. No matter who is right ( although I am leaning towards the conclusion by Brett et.al., partly because I am amazed by DSCAM), I totally agree with the following argument put forward by Bork and colleagues:
"...no estimate should be considered absolute when extrapolating from data with many hidden biases...the art is to avoid hidden biases in derived data and their processing as much as possible."Posted by yxing at October 12, 2004 10:32 AM
Pretty good,sometimes I really think that bioinformatics is a subject digging data on data,while the later one many times are not solid enough,how can the former be?
tons of bioinformatics papers will analyze the corresponding positve and negative data as much as possible.Well, the same consideration already there:)
Great !
Posted by: mitsuteru_nakao at October 13, 2004 01:17 AMActurally, I don't like the word "bioinformatics". I and my colleagues prefer to use "Computational Biology". I think it's a misunderstand that the "damn prediction" is our mainly aim. But, I think we should focus on elaborated analysis, which use computer as a tool, not aim.
Posted by: lylover at October 13, 2004 11:08 PMOnce I think about this question:
Human has 32,000 genes, but plant has >twice (not Arabidopsis, but such as rice). Why?
Maybe, AS is the key. Animal prefer to use AS to raise the complexity of genome, but plant use duplicated genes more. Then...
Posted by: lylover at October 13, 2004 11:11 PM