There’s a very interesting article that came out recently describing the first public report of carbapenem resistance in E. coli from the U.S. food supply. Maryn McKenna provides a good overview:

Bacteria containing a gene that confers resistance to a crucial class of antibiotics have been found in buildings on a pig farm in the midwestern United States, a troubling and mysterious discovery that should ring a warning bell over farm antibiotic use.

The discovery is troubling because the drugs to which the bacteria are resistant are so important. Called carbapenems, they are one of the few drug families that still work against deadly, multi-drug-resistant hospital infections. Carbapenem-resistant bacteria, known as CREs, are considered an “urgent threat” by the Centers for Disease Control and Prevention.

The discovery is mysterious because carbapenems are not used on that farm; in fact, they are not used in agriculture anywhere in the world. So the resistance may have been transported onto the farm in a manner no one can yet explain.

But there’s no question that the discovery is alarming. It’s likely to bring fresh scrutiny to the use of antibiotics on farms in the US, just as a long-delayed set of FDA restrictions on farm use is finally clicking into place.

But that’s not what this post is about. What this post is about is how the authors sequenced the plasmid (a mini-chromosome that can jump from bacterium to bacterium) that contains the carbapenemase gene, and then didn’t deposit the sequence in Genbank (at least as of mid-day Dec. 8th). Mind you, Genbank deposition is standard procedure for most journals (and federally funded work), including the journal in which the study was published. There is no Genbank accession listed–this isn’t a case of a sequence being held by NCBI until the paper is ‘officially published’. The authors were able to put together a map of the plasmid for the paper, so it’s clear there is a sequence:

Other researchers who could use these data for their own work (and verify the validity of the authors’ results as well) are unable to do so, and, as McKenna notes, this is a critical health problem. The worst case scenario–one I think is highly unlikely–is that they could have interpreted the data incorrectly and carbapenemase resistance is not due to the putative carbapenemase gene (e.g., there is a stop codon in the middle of the gene, gene isn’t full length, etc.). Reproducibility requires, at the very least, making the sequence publicly available.

The editors and the reviewers should have caught this. It’s unfortunate because these data could help many researchers, including public health professionals. As genomics becomes more ‘democratized’, it’s all the more vital that the widely accepted standard of post-publication data sharing is enforced by editors and reviewers.

Hopefully, there will be a clarification and correction soon.