A project-selection audit of public urinary-source Escherichia coli genomes from China distinguishes purpose-enriched NDM and mcr signals from CTX-M recurrence across projects

Bacteria are constantly evolving, and one of their most dangerous tricks is learning to survive the medicines we use to kill them. When a bacterium like Escherichia coli becomes resistant to antibiotics, it can turn a common infection, such as a urinary tract infection, into a life-threatening condition. To track this threat, scientists around the world have been sequencing the genetic code of these bacteria and storing the results in massive public archives. Think of these archives as a giant library of bacterial blueprints, where researchers can look up the genetic history of infections from different places and times. However, just like any library, the collection is not a perfect random sample of everything that exists. Some books are there because they are rare and interesting, while others are there because they represent a typical day. If a researcher looks only at the rare books, they might think the world is full of monsters, when in reality, most of the library contains ordinary stories. Understanding the difference between a rare, special case and a common, recurring pattern is essential for knowing how dangerous a specific type of bacteria really is.

A team of researchers from Ningbo Medical Center Lihuili Hospital decided to take a close look at the public library of E. coli genomes from China to see how the collection was put together. They gathered 512 bacterial genomes that came from urine samples, a common source of infection. Instead of just counting how many bacteria had specific resistance genes, they first investigated the "story" behind each batch of data. They asked a simple but crucial question: Why was this specific group of bacteria chosen to be sequenced? Was it because the researchers were looking for a general sample of patients, or were they specifically hunting for bacteria that had already shown signs of being resistant to powerful drugs? By tracing the original research papers and project records for each group, they found that the library was heavily skewed. Nearly seventy percent of the bacteria in the collection came from studies that were specifically designed to find resistant strains. These were not random snapshots of the population; they were targeted searches for the toughest bugs.

This distinction turned out to be the key to unlocking the true picture of resistance. The researchers found that the most dangerous resistance genes, those that allow bacteria to survive carbapenems (a class of last-resort antibiotics) and colistin (another powerful drug), were almost entirely concentrated in these targeted studies. Every single bacterium in the collection that carried the gene for carbapenem resistance came from a project that was explicitly looking for such resistance. Similarly, nearly all the bacteria carrying the colistin-resistance gene came from these same specialized collections. In contrast, a different type of resistance gene, which makes bacteria resistant to a common group of antibiotics called cephalosporins, appeared in almost every project, including those that were just looking at general patient samples. This gene was found in nearly half of the bacteria from the general collections and remained common even when the researchers removed the largest, most targeted studies from their analysis. This told the scientists that while the super-resistant genes were rare and concentrated in specific research efforts, the cephalosporin-resistance gene was a widespread, recurring problem that existed across the entire population of bacteria, regardless of how the data was collected.

The study also looked at the family trees of these bacteria to see if certain lineages were more likely to carry these dangerous genes. They found that specific genetic families, known as sequence types, had very different personalities. Two families, ST410 and ST167, were the primary hosts for the carbapenem-resistance genes, acting as the main carriers for these super-resistant traits. Another two families, ST131 and ST1193, were more associated with resistance to fluoroquinolones and the widespread cephalosporin-resistance gene, but they rarely carried the carbapenem-resistance genes. The researchers also examined the genetic neighborhoods where these resistance genes lived. They discovered that the carbapenem-resistance gene was usually found on a specific type of mobile genetic vehicle, similar to a plasmid, that acts like a delivery truck for resistance. The colistin-resistance gene was found on different delivery trucks. This helped them understand that these dangerous traits were not just floating randomly but were hitching rides on specific genetic structures that moved between bacteria.

The researchers concluded that the way we look at public data matters just as much as the data itself. If you simply count the number of resistant bacteria in a public archive without checking why they were collected, you might get a distorted view of the threat. The archive is full of concentrated signals of the most dangerous resistance because scientists have been actively hunting for them, but it also contains a steady, recurring signal of more common resistance that exists in the general population. By auditing the purpose of each study and combining that with a detailed look at the bacterial genetics, the team was able to separate the rare, concentrated threats from the common, widespread ones. This approach helps scientists understand the true composition of the bacterial world, ensuring that when they see a pattern in the data, they know whether it is a sign of a widespread epidemic or simply the result of a very focused search.