5
$\begingroup$

Despite the fact that the human genome project was declared "complete" in 2001, there are even now still gaps due to difficult-to-sequence regions of the genome such as telomeres and centromeres. Only last year was the first entirely complete chromosome sequence produced.

Telomeres and centromeres, however, are specific to eukaryotes. Are there similar challenges with prokaryotes? A recent publication on a gapless E. coli sequence implies that there are indeed typically gaps, but I can't tell whether this indicates any systematic challenges, as with the human genome.

Thus, my question: are there known genome sequencing issues that currently make it difficult to acquire complete genome sequences from common lab strains such as E. coli?

$\endgroup$

2 Answers 2

3
$\begingroup$

Generally, gaps in genomes occur due to failure of de novo assembly -- the DNA can be sequenced accurately, but cannot subsequently be placed into a contiguous assembly due to alignment ambiguity with adjacent sequences. In eukaryotes, telomeres and centromeres are difficult to assemble from short reads because they are highly repetitive. In bacteria, repetitive regions of the genome can cause similar problems when assembling from short reads. Such regions include CRISPR arrays, integrated mobile elements (e.g. transposons), and gene duplications.1,2

So, what made a short-read no-gap assembly of E. coli strain C difficult, specifically? Alignment of the original short-read assembly 3 to the long-read reference 4 with Mauve 5 gives positions and length of gaps.

mauve whole-genome alignment of E. coli C genomes from short and long reads

The first gap in the ordered short-read assembly (left-most red line in the figure) occurs at position 15,085 in the first contig and is 320 bp long. Looking at the same region in the long-read assembly, this closely corresponds to a 236 bp tandem repeat (CTGG × 59) at position 14,848. The authors of the short-read paper used 150 bp paired-end MiSeq reads to create their assembly, which, after adapter trimming, would likely not give enough sequence context flanking the repeat region to span it with a single read, thus creating a gap. This is just one example, and the rest of the gaps could also be repeats or other aforementioned mobile elements.


  1. Treangen TJ, Abraham AL, Touchon M, Rocha EP. Genesis, effects and fates of repeats in prokaryotic genomes. FEMS Microbiol Rev. 2009 May;33(3):539-71.
  2. Ricker N, Qian H, Fulthorpe RR. The limitations of draft assemblies for understanding prokaryotic adaptation and evolution. Genomics. 2012 Sep;100(3):167-75.
  3. https://www.ncbi.nlm.nih.gov/nuccore/MNKV00000000
  4. https://www.ncbi.nlm.nih.gov/nuccore/CP029371
  5. Darling AC, Mau B, Blattner FR, Perna NT. Mauve: multiple alignment of conserved genomic sequence with rearrangements. Genome Res. 2004 Jul;14(7):1394-403.
$\endgroup$
4
  • $\begingroup$ Does this mean then that long-read sequencing, like with PacBio or Oxford Nanopore, shouldn't expect to run into any gap issues in E. coli and similar? $\endgroup$
    – jakebeal
    Mar 8, 2021 at 16:14
  • $\begingroup$ The key point is that many regions that cause short-read assemblers to fail (causing gaps) can be spanned by long reads. Repeated regions and tandem repeats in bacteria tend to be on the order of hundreds or thousands of bp long, and can therefore be captured by single long reads, which can be >10k bp for Nanopore and >100k bp for PacBio. $\endgroup$
    – acvill
    Mar 8, 2021 at 16:24
  • $\begingroup$ Thank you, this is very clear @acvill. Out of curiosity, are the eukaryotic repeats much longer and thus harder to deal with, or are those readily addressed by recent sequencing technology as well? $\endgroup$
    – jakebeal
    Mar 8, 2021 at 16:28
  • 1
    $\begingroup$ I know less about eukaryotic repeats, but the background of this paper seems to be a good resource: genomebiology.biomedcentral.com/articles/10.1186/…. If you have more questions, I'd encourage you to make a new post. $\endgroup$
    – acvill
    Mar 8, 2021 at 16:42
2
$\begingroup$

Summary

The publication of the first complete DNA sequence of Escherichia coli was for strain K-12 substr. MG1655 in 1997. Because of historical limitations in the technology used (rather than problems with repeat sequences or instability in the genome) this contained mistakes, which were subsequently corrected as the sequencing technology improved. An international consortium reported on this (and revisions in gene annotation) in a paper in 2006. Because of the extensive use of E. coli in biological research in the 20th century, many different strains have been used for different purposes and many sequences have been reported. The strain referred to in the question is strain C, which has only just been completely sequenced (or its sequence only just reported), presumably because nobody had the need to do so previously.

Technological context

The Wikipedia entry on DNA Sequencing contains the graphic reproduced below, showing the decline in the cost of sequencing between 2001 and 2019.

Decline in cost of DNA sequencing

[Based on data from the National Human Genome Research Institute]

This decline in cost was the result of both improvement in methodology and automation, and also reflects the speed with which sequences can be obtained and hence the ease of repeating them. The E. coli DNA sequence published in 1997, and was only the seventh complete genome to be reported. The following extracts from the section of the paper on sequencing strategy gives an idea of the the various methodologies employed and problems encountered:

Sequencing was carried out in sections, with steadily improving technical approaches. The M13 Janus shotgun strategy proved to be the most efficient strategy for data collection and closure… The first 1.92 Mb… was sequenced from our overlapping set of 15- to 20-kb MG1655 lambda clones by means of radioactive chemistry and was deposited in GenBank between 1992 and 1995. Subsequently, we switched to dye-terminator fluorescence sequencing (Applied Biosystems). In addition to greater speed and lower cost, this new technology avoided electrophoretic compression artifacts, which, owing to its 50.8% G􏰂C content, occur in practically every gene of E. coli…The final stages entailed special attention to problem areas. The region between positions 0 and 22,551 did not yield a suitable I–Sce I fragment, so three lambda clones were selected to finally complete the genome. One of them was found to contain a deletion and had to be finished by shotgun sequencing of a long-range polymerase chain reaction (PCR) fragment.

Note in particular that newer, more accurate, sequences were combined with individual genes previously sequenced, and the reference to compression artefacts — the bugbear of separation of fragments by polyacrylamide gel electrophoresis. Not mentioned is the mistakes that can be made using the Maxam–Gilbert (chemical) method if one is not aware of the effect of methylation.

It is not surprising that there were mistakes in this sequence (and, indeed, in many much shorter sequences, often published in a hurry to obtain priority.)

Tracing the history of the K12 sequence

I think it worth explaining how I found out about the subsequent changes in the sequence, as the general approach may be useful to those that are not aware of it.

I started by locating the original paper by a simple Internet search. From this I obtained the accession number (an ID) of its deposition at GenBank — AE000137 — and searched for this on the NCBI website and found that it had been replaced by another entry, U00096. In the header of the entry, after the summary information regarding the organism and sequence, there is a reference section of the type:

REFERENCE   1  (bases 1 to 4641652)
  AUTHORS   Blattner,F.R., Plunkett,G. III, Bloch,C.A., Perna,N.T., Burland,V.,
            Riley,M., Collado-Vides,J., Glasner,J.D., Rode,C.K., Mayhew,G.F.,
            Gregor,J., Davis,N.W., Kirkpatrick,H.A., Goeden,M.A., Rose,D.J.,
            Mau,B. and Shao,Y.
  TITLE     The complete genome sequence of Escherichia coli K-12
  JOURNAL   Science 277 (5331), 1453-1462 (1997)
   PUBMED   9278503
REFERENCE   2  (bases 1 to 4641652)
  AUTHORS   Hayashi,K., Morooka,N., Yamamoto,Y., Fujita,K., Isono,K., Choi,S.,
            Ohtsubo,E., Baba,T., Wanner,B.L., Mori,H. and Horiuchi,T.
  TITLE     Highly accurate genome sequences of Escherichia coli K-12 strains
            MG1655 and W3110
  JOURNAL   Mol. Syst. Biol. 2, 2006 (2006)
   PUBMED   16738553
REFERENCE   3  (bases 1 to 4641652)
  AUTHORS   Riley,M., Abe,T., Arnaud,M.B., Berlyn,M.K., Blattner,F.R.,
            Chaudhuri,R.R., Glasner,J.D., Horiuchi,T., Keseler,I.M., Kosuge,T.,
            Mori,H., Perna,N.T., Plunkett,G. III, Rudd,K.E., Serres,M.H.,
            Thomas,G.H., Thomson,N.R., Wishart,D. and Wanner,B.L.
  TITLE     Escherichia coli K-12: a cooperatively developed annotation
            snapshot--2005
  JOURNAL   Nucleic Acids Res. 34 (1), 1-9 (2006)
   PUBMED   16397293
  REMARK    Publication Status: Online-Only

These three are journal references: there are 15 others which are just direct submissions to GenBank, most by Blattner’s group. Some of the earlier ones are marked as corrections. The later ones I assume are mainly concerned with the annotation of genes rather than the sequence itself. The latest date is 2014.

The authors of the 2006 Nucleic Acid Research paper write:

“Corrections to the original MG1655 genome are at 243 sites (totaling 358 nt), a correction rate 8 years later of 􏰍~7 in 105.”

They also make clear the problems of differences in what purport to be the same strains, which have mutated over the years through passaging and transfer to different labs (despite, no doubt, the best efforts to control this).

Reporting bacterial sequences today

Even in 2006 it required some effort to collaborate and correct a bacterial sequence, whereas today (2021) it would appear trivial to resequence any strain of interest. Thousands of E. coli sequences must have been determined. It may be of practical importance to workers using different bacterial strains to know when they have been sequenced, but it is now impossible to publish such in a standard scientific journal. It appears that the American Society for Microbiology has addressed that problem in a section of their journal (Microbiology) entitled “Resource Announcements”, which are little more than announcements of the deposition of a particular sequence. This would be the case for E. coli, strain C, which provoked this question.

$\endgroup$
4
  • $\begingroup$ I'm afraid that, while this is a nice summary of historical challenges, it doesn't really address the current state of play. $\endgroup$
    – jakebeal
    Mar 9, 2021 at 10:54
  • 1
    $\begingroup$ With all due respect, your question was about whether there were general problems sequencing E. coli and why a recent report made a point of the fact that the sequence was ungapped. I don't see anything about the current "state of play" in the question. It seemed necessary to explain why that the gaps in E. coli sequences were due to deficiencies in the technology when they performed, rather than structural problems, as in telomeres. And rather than just telling you this, I explained how to find this out for yourself, as your question seemed to indicate an ignorance of sequencing history. $\endgroup$
    – David
    Mar 9, 2021 at 12:23
  • $\begingroup$ Apologies; I thought that was clear from my discussion of systematic challenges and the nature of the organism, as opposed to technological challenges. I've added the word "current" to make it explicit. $\endgroup$
    – jakebeal
    Mar 9, 2021 at 12:32
  • $\begingroup$ @jakebeal Whatever. I try to make answers generally useful, rather than merely to the poster. I think this one is, as ignorance of the development of science and how to use GenBank entries is widespread. $\endgroup$
    – David
    Mar 9, 2021 at 12:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .