Genome data may contain small but significant errors

Since the genome sequence of the bacterial pathogen Haemophilus influenzae was published in 1995, the genetic code of many other large, complex, medically, and commercially significant organisms including humans has also been elucidated.

However, the techniques used to derive these genetic sequences are imperfect, and many researchers may be unaware of potential errors lurking within the publicly available published, or "canonical" sequence. If an organism's genome is unstable, variable, and contains rearrangements within a population or between strains, there may be no single true linear structure that will be valid for that organism, and imposing a linear sequence may not be biologically meaningful.

Now, researchers at Cold Spring Harbor Laboratory and New York University describe a high throughput microarray technique that involves testing many samples simultaneously and which can be used to assemble physical maps and validate genomic sequence assemblies. The findings appear in the latest issue of the Journal of Computational Biology.

The research was conducted by Joseph West, John Healy, and Michael Wigler of Cold Spring Harbor Laboratory, and William Casey and Bud Mishra of NYU's Courant Institute of Mathematical Sciences. Mishra is a Professor of Computer Science and Mathematics at the Courant Institute and also has an appointment in the Department of Cell Biology at NYU's School of Medicine.

Using their micro-array hybridization method, which used flourescently labeled snippets from the genome of the fission yeast S. pombe and examined how they bind to probes arrayed on a glass slide, they were able to computationally derive the "distance" between probes in the genome and organize the probes along the genome. The resulting physical map of the S. pombe genome was compared to the corresponding map computed from publicly available S. pombe sequence. The comparison showed a small number of significant discrepancies between their results and that of the map derived from the public sequence released in 2002. S. pombe's genome is only about 14 million bases long (almost a thousandth of the human genome), and is widely considered to be a gold-standard in whole-genome assembly.

The authors show that with appropriate experimental conditions, array hybridization data can be used to establish a physical distance between unique arrayed probes--a sequence of DNA which in this case was 70 base pairs long. Each of the 70 base pairs is unique in the target organism's genome and serves as a landmark in that genome. These probes can then be ordered in the correct sequence in which they occur in the target genome in much the same way as a mapmaker can locate landmarks at the correct coordinates by consulting a three-dimensional rendering of a geographical map. The distance between pairs of landmarks can be used to assemble physical maps, as an aid to sequence assembly, or as an independent method for validating sequence assembly and indicating where errors need correction.

Comparing data from their inferred probe maps to the available sequence assembly, the new method provides insights into the difficulties of establishing a canonical and accurate sequence or physical map, and suggests ways that the two types of data can be combined to render increased confidence levels of the assembly.

This physical mapping technology is simple to implement and is relatively inexpensive. It is likely to have significant commercial impact through disease-related genetics studies, such as cancer and autism. In addition, it complements other mapping and sequencing technologies (e.g., Optical Mapping and Sequencing being developed by Mishra) and cancer array CGH studies (e.g., ROMA project of Wigler and a versatile cancer genome analysis project of Mishra).