Saturday, February 06, 2021

The 20th anniversary of the human genome sequence:
2. Finishing the sequence

It's been 20 years since the first drafts of the human genome sequence were published. These first drafts from the International Human Genome Project (IHGP) and Celera were far from complete. The IHGP sequence covered about 82% of the genome and it contained about 250,000 gaps and millions of sequencing errors.

Celera never published an updated sequences but IHPG published a "finished" sequence in October 2004. It covered about 92% of the genome and had "only" 300 gaps. The error rate of the finished sequence was down to 10-5.

International Human Genome Sequencing Consortium (2004) Finishing the euchromatic sequence of the human genome. Nature 431:931-945. doi: 10.1038/nature03001

We've known for many decades that the correct size of the human genome is close to 3,200,000 kb or 3.2 Gb. There's isn't a more precise number because different individuals have different amounts of DNA. The best average estimate was 3,286 Gb based on the sequence of 22 autosomes, one X chromosome, and one Y chromosome (Morton 1991). The amount of actual nucleotide sequence in the latest version of the reference genome (GRCh38.p13) is 3,110,748,599 bp and the estimated total size is 3,272,116,950 bp based on estimating the size of the remaining gaps. This means that 95% of the genome has been sequenced. [see How much of the human genome has been sequenced? for a discussion of what's missing.]

Recent advances in sequencing technology have produced sequence data covering the repetitive regions in the gaps and the first complete sequence of a human chromosome (X) was published in 2019 [First complete sequence of a human chromosome]. It's now possible to complete the human genome reference sequence by sequencing at least one individual but I'm not sure that the effort and the expense are worth it.


Image credit the figure is from Miga et al. (2019)

Miga, K.H., Koren, S., Rhie, A., Vollger, M.R., Gershman, A., Bzikadze, A., Brooks, S., Howe, E., Porubsky, D., Logsdon, G.A. et al. (2019) Telomere-to-telomere assembly of a complete human X chromosome. Nature 585:79-84. [doi: 10.1038/s41586-020-2547-7]

Morton, N.E. (1991) Parameters of the human genome. Proceedings of the National Academy of Sciences 88:7474-7476. [doi: 10.1073/pnas.88.17.7474]

10 comments:

  1. As I understand it the 5% that is remaining as unsequenced gaps are difficult to sequence primarily because they're highly repetitive in nature, is that correct?
    If so, can you say more about why sequencing repetitive DNA is technically challenging?

    ReplyDelete
    Replies
    1. .. oh and if that isn't what makes it difficult, then what is?

      Delete
    2. I think Larry is saying that it's also due to regions in which there are deletions and insertions that vary from individual to individua. So even if we could perfectly sequence two genomes, their total lengths would not be the same. We can get closer and closer to knowing the average length, but that's nit the sane as the length. It us like having a massive project to determine the height of the human being, by measuring one human more and more precisely.

      Delete
    3. I would suppose that it's not the sequencing that's difficult, it's the assembly. If a repetitive region is significantly longer than the length of the typical read, it's hard to concatenate those reads into a complete sequence.

      Delete
    4. One of the main problems with the earliest sequencing/assembly techniques is that the DNA being sequenced had to be cloned and propagated in bacteria. Highly repetitive regions are unstable in plasmid/cosmids/BACs so their length and sequences couldn't be accurately determined.

      In addition, it's difficult to sequence highly repetitive DNA because of stuttering and stammering during the DNA polymerase reaction. The third problem has to do with assembly, as John Harshman noted. Even if you have accurate reads of 2000 bp there are many possible ways of assembling these reads into a contiguous region. This has been particularly difficult in centromeric regions but with the development of long sequence reads this problem has been overcome.

      Delete
    5. My experience with Sanger sequencing is that any single-base repeats of 10 or more result in an unreadable binomial distribution of repeat lengths that makes all the downstream sequence unreadable. Or maybe that's the prior PCR, because cloning seems to prevent the problem.

      Delete
  2. My question is: Why not say billion bases, rather than Gigabases?

    ReplyDelete
    Replies
    1. Go ahead and say "billion bases" if you want. Gigabase, megabase, kilobase, etc. follow the standard metric naming conventions and have simple, widely understood abbreviations (Gb, Mb, Kb, etc.).

      Delete