Tales of drafty genomes: part 1 — The Human Genome

One of my recent blog posts discussed this new paper in PLOS Computational Biology:

There has also been a lot of chatter on twitter about this paper. Here is just part of an exchange that I was involved in yesterday:


The issue of what is or isn’t a draft genome — and whether this even matters — is something on which I have much to say. It’s worth mentioning that there are a lot of draft genomes out there: Google Scholar reports that there are 1,440 artices that mention the phrase ‘draft genome’ in their title [1]. In the first part of this series, I’ll take a look at one of the most well-studied genome sequences in existence…the human genome.

The most famous example of a draft genome is probably the ‘working draft’ of the human genome that was announced — with much fanfare — in July 2000 [2]. At this time, the assembly was reported as consisting of “overlapping fragments covering 97 percent of the human genome”. By the time the working draft was formally published in Nature in January 2001, the assembly was reported as covering “about 94% of the human genome” (incidentally, this Nature paper seems to be first published use of the N50 statistic).

On April 14, 2003 the National Human Genome Research Institute and the Department of Energy announced the “successful completion of the Human Genome Project” (emphasis mine). This was followed by the October 2004 Nature paper that discussed the ongoing work in finishing the euchromatic portion of the human genome[3]. Now, the genome was being referred to as ‘near-complete’ and if you focus on the euchromatic portion, it was indeed about 99% complete. However, if you look at the genome as a whole, it was still only 93.5% complete [4].

Of course the work to correctly sequence, assemble, and annotate the human genome has never stopped, and probably will never stop for some time yet. As of October 14, 2014, the latest version of the human genome reference sequence is GRCh38.p1[5] lovingly maintained by the Genome Reference Consortium (GRC). The size of the human genome has increased just a little bit compared to the earlier publications from a decade ago[6], but there is still several things that we don’t know about this ‘complete/near-complete/finished’ genome. Unknown bases still account for 5% of the total size (that’s over 150 million bp). Furtheremore, there are almost 11 million bp of unplaced scaffolds that are still waiting to be given a (chromosomal) home. Finally, there remains 875 gaps in the genome (526 are spanned gaps and 349 unspanned gaps[7]).

If we leave aside other problematic issues in deciding what a reference genome actually is, and what it should contain[8], we can ask the simple question is the current human genome a draft genome? Clearly I think everyone would say ‘no’. But what if I asked is the current human genome complete? I’m curious how many people would say ‘yes’ and how many people would ask me to first define ‘complete’.

Here are some results for how many hits you get when Googling for the following phrases:

Scientists and journalists don’t help the situation by maybe being too eager to overhype the state of completion of the human genome[9]. In conclusion, the human genome is no longer a draft genome, but it is still just a little bit drafty. More on this topic of drafty genomes in part 2!

  1. There are 1,570 if you don’t require the words ‘draft’ and ‘genome’ to be together in the article title.  ↩

  2. The use of the ‘working draft’ as a phrase had been in use since at least late 1998.  ↩

  3. There is also the entire batch of chromosome-specific papers published between 2001 and 2006.  ↩

  4. This percentage is based on the following line in the paper: “The euchromatic genome is thus ~2.88 Gb and the overall human genome is ~3.08 Gb”  ↩

  5. This is the 1st patched updated to version 38 of the reference sequence  ↩

  6. There are 3,212,670,709 bp in the latest assembly  ↩

  7. The GRC defines the two categories as follows:

    Spanned gaps are found within scaffolds and there is some evidence suggesting linkage between the two sequences flanking the gap. Unspanned gaps are found between scaffolds and there is no evidence of linkage.  ↩

  8. Remember, human genomes are diploid and not only vary between individuals but can also vary from cell-to-cell. The idea of a ‘reference’ sequence is therefore a nebulous one. How much known variation do you try to represent (the GRC represents many alternative loci)? How should a reference sequence represent things like ribosomal DNA arrays or other tandem repeats?  ↩

  9. Jonathan Eisen wrote a great blog post on this: Some history of hype regarding the human genome project and genomics  ↩

Recent changes to the ACGT blog

There's probably some published rules for how to write blog posts and I imagine that one of those rules might be don't blog about your blog. Oh well…

Over the last few months I've been making lots of tweaks to this site. Most of them have been subtle changes in order to make the pages less cluttered and more aesthetically pleasing — e.g. did you notice that I moved the 'About Blog Contact' menu bar ~20 pixels closer to the top of the page as it was just a little bit too lonely where it used to be? Aside from these minor cosmetic alterations, I've also made some more substantial changes:

  • You can no longer see tags for individual blog posts (but I have kept the tag cloud on the About page so you can still find all posts that share the same tag).
  • On the same About page, I added an option to let people subscribe to the blog via email (maximum one email per day).
  • Just about every post listed on the main page of this site used to have a 'Read More' link which you needed to click in order to read the entire article. I'm now restricting this only to my longer posts.
  • I've disabled comments from the blog; partly because I wasn't getting a lot of them but mostly because of the many excellent reasons stated on the @avoidcomments Twitter account.
  • I've started including occasional 'link blog' posts. These are short posts — often just a paragraph or two — that typically comment on a paper or on another blog post. The titles of link blog posts are themselves links to the source paper/blog post, and include a little arrow to indicate this. E.g. see here, here, or here.

As a final note, I'd like to thank everyone who has tweeted or otherwise spread the word about this blog. Aside from churning out more posts about bioinformatics tools with bogus names, I do want to make a bit more of an effort to write about some more substantive issues in bioinformatics. Stay tuned!

Ten recommendations for software engineering in research →

The list of recommendations in this new GigaScience paper by Janna Hastings et al. is not aimed at bioinformatics in particular, but many bioinformaticians would benefit from reading it. I can particularly relate to this suggestion:

Document everything
Comprehensive documentation helps other developers who may take over your code, and will also help you in the future. Use code comments for in-line documentation, especially for any technically chal- lenging blocks, and public interface methods. However, there is no need for comments that mirror the exact detail of code line-by-line.

3 word summary of new PLOS Computational Biology paper: genome assemblies suck

Okay, I guess a more accurate four word summary would be 'genome assemblies sometimes suck'.

The paper that I'm referring to, by Denton et al, looks at the problem of fragmentation (of genes) in many draft genome assemblies:

As they note in their introduction:

Our results suggest that low-quality assemblies can result in huge numbers of both added and missing genes, and that most of the additional genes are due to genome fragmentation (“cleaved” gene models).

One section of this paper looks at the quality of different versions of the chicken genome and CEGMA is one of the tools they use in this analysis. I am a co-author of CEGMA, and reading this paper brought back some memories of when we were also looking at similar issues.

In our 2nd CEGMA paper we tried to find out why 36 core genes were not present in the v2.1 version of the chicken genome (6.6x coverage). Turns out that there were ESTs available for 29 of those genes, indicating that they are not absent from the genome, just from the genome assembly. This led us to find pieces of these missing genes in the unanchored set of sequences that were included as part of the set of genome sequences (these often appear as a 'ChrUn' FASTA file in genome sequence releases).

Something else that we reported on in our CEGMA paper is that, sometimes, a newer version of a genome assembly can actually be worse than what it replaces (at least in terms of genic content). Version 1.95 of the Ciona intestinalis genome contained several core genes that subsequently disappeared in the v2.0 release.

In conclusion — and echoing some of the findings of this new paper by Denton et al.:

  1. Many genomes are poorly assembled
  2. Many genomes are poorly annotated (often a result of the poor assembly)
  3. Newer versions of genome assemblies are not always better