Friday, February 08, 2008

Junk in Your Genome: Protein-Encoding Genes

The typical human gene has eight exons and seven introns (the actual average number of introns is 7.2). These values are based on analysis of 5236 well-characterized human genes with full-length cDNA's (Hong et al. 2006). There are lots of conflicting results in the literature. Most claim there are more introns but the data is based largely on a computational assessment of introns and exons. It includes a number of introns of extraordinary length lying between exons of dubious existence (often non-coding). I'll assume for the time being that there are 7.2 introns per gene, on average, and the average length is 3750 bp (Hong et al. 2006)

Each gene is transcribed from a 5′ promoter (P) and the primary transcript terminates at a polyadenylation site (t).

THEME

Genomes & Junk DNA

Total Junk so far

    55%
The exons contain coding regions (blue) that encode the sequence of the protein product. A typical protein has a molecular weight of 70,000 daltons and this corresponds to about 635 amino acid residues. The coding region is 1905 bp but we'll round up to 2 kb. Each gene has a region of the mRNA at the 5′ end called the 5′ untranslated region (UTR). This is required for translation. It averages 200 bp in size, with considerable variation. The 3′ end of the gene has a similar untranslated region that we'll assume to be essential.

Thus, total essential exons comprise 2200 bp on average per gene. Since there are 20,500 protein-encoding genes, this means 20,500 × 2.2 kb = 45.1 Mb or 1.4% of the genome (about 1.3% coding and 0.1% UTRs).

The minimum size of a eukaryotic intron is less than 50 bp. For a typical mammalian intron, the essential sequences in the introns are: the 5′ splice site (~10 bp); the 3′ splice site (~30 bp): the branch site (~10 bp); and enough additional RNA to form a loop (~30 bp). This gives a total of 80 bp of essential sequence per intron or 20,500 × 7.2 × 80 = 11.8 Mb. Thus, 0.37% of the genome is essential because it contains sequences for processing RNA.

The total of essential sequences in the transcribed part of a gene is about 1.8% of the genome.

The rest of the intron sequence is non-essential junk. Much of it is littered with transposable elements that have inserted haphazardly. If we subtract the essential intron sequence then the average size of the remaining DNA is 3650 bp. The total amount of this sequence is 20,500 × 7.2 × 3650 = 538.7 Mb or 17% of the genome. (Most estimates are somewhat higher.)

Assuming that 44% of this is repetitive transposable elements, this leaves 7.4% 9.6% of the genome. That's an additional 7.4% 9.6% of non-essential DNA, or junk, bringing our current total to 53% 55% junk.

The transcription of every gene is controlled by sequences beyond the 5′ end. There are two classes of sequence; promoters, and regulatory sequences. The actual binding sites for RNA polymerase II and various regulatory proteins make up only about 100 bp of essential sequence but the various bound proteins have to form loops of DNA in order to come into contact. It's reasonable to assume that the average gene may need as much as 1000 bp of essential regulatory sequence. (A generous estimate.)

This means 20,500 × 1000 bp = 20.5 Mb or 0.6% of the genome is essential for regulation.

The grand totals for protein-encoding genes are:

essential 2.4%

junk 7.4% 9.6% (not counting sequences that were included in other calculations)


Hong X, Scofield DG, Lynch M (2006) Intron size, abundance, and distribution within untranslated regions of genes. Mol. Biol. Evol. 23:2392-404. [PubMed]

11 comments:

  1. I don't think it would make any significant impact on your calculations but there are microRNAs encoded within introns, intronic microRNA. I wonder if there is any other essential noncoding RNA that you might miss in your calculations.

    ReplyDelete
  2. Thus, total essential exons comprise 2200 bp on average per gene. Since there are 20,500 protein-encoding genes, this means 20,500 × 2.2 kb = 45.1 Mb or 1.4% of the genome (about 13% coding and 1% UTRs).
    Shouldn't that be
    1.3% coding and 0.1% UTRs?

    ReplyDelete
  3. rob,

    I don't think it would make any significant impact on your calculations but there are microRNAs encoded within introns, intronic microRNA. I wonder if there is any other essential noncoding RNA that you might miss in your calculations.

    I haven't covered small RNAs yet. Wait for it.

    I also haven't covered all the essential non-transcribed sequences in "noncoding" DNA.

    ReplyDelete
  4. soparc asks,

    Shouldn't that be
    1.3% coding and 0.1% UTRs?


    Yes it should. Thanks.

    ReplyDelete
  5. Do you believe that the "typical human exon" is about 250 bp in size?

    You're saying there 8.2 exons on average, and 2,200 bp of mRNA, which would imply an average exon size of ~250 bp - about 100 bp more than what I've read in several papers.


    (sorry if my comment shows up twice, I tried posting a few hours but it hasn't shown up, while your replies to previous comments have - hence this second attempt)

    ReplyDelete
  6. rileen asks,

    Do you believe that the "typical human exon" is about 250 bp in size?

    You're saying there 8.2 exons on average, and 2,200 bp of mRNA, which would imply an average exon size of ~250 bp - about 100 bp more than what I've read in several papers.


    The short answer is "yes," I do believe that the average exon is about 250 bp.

    The long answer requires an extended discussion about the reliability of EST databases and the predictions of alternative splicing. That's why I choose to use values from well-characterized genes where real live human scientists have looked at the data and weeded out the nonsense.

    ReplyDelete
  7. Fair enough - as student pursuing a Ph.D in Bioinformatics, and studying Alternative Splicing in particular, I find your posts very interesting, and something that helps retain some sort of "perspective" via exposure to the other extreme of opinion, so to say.

    ReplyDelete
  8. Larry, if the amount of necessary sequences within introns are as small as you suggest wouldn't this this allow us to make a prediction. Couldn't we predict that due to drift there should be very little similarity in intron lengths between different species. If, by any chance, there is similarity then what would your explanation be?

    ReplyDelete
  9. If stuff is either "essential" or "junk", you might as well trim off all those junky crystallographically-unstructured loop domains from the coding sequences of protein-encoding exons too. Out of 635 amino acids in the average protein, you could probably toss out a third of those, maybe more.

    ReplyDelete
  10. If stuff is either "essential" or "junk", you might as well trim off all those junky crystallographically-unstructured loop domains from the coding sequences of protein-encoding exons too

    Not really, depending on the protein that unstructured loop actually has functions, typically allowing protein domains to move quite bit.

    ReplyDelete
  11. It includes a number of introns of extraordinary length lying between exons of dubious existence (often non-coding).

    A non-protein-coding exon is still an exon. Many 5' exons are not translated, but are still functional sequences (3'UTRs). I agree that computationally predicted 5' exons are often errors, but that doesn't mean that some 5' exons are important. Also, those introns of extraordinary length often contain important cis regulatory sequences (long introns are under more selective constraint than short introns).

    ReplyDelete