NEW YORK, Nov. 21 - All mirror sites tied to the GDB human genome database have regained their ability to access updated research data from the primary site, GenomeWeb has learned.
Connie Talbot, who curates the GDB website from Johns Hopkins University, confirmed on Wednesday that "the [mirror] nodes are getting their feeds and the missing data are there."
GenomeWeb originally learned that the mirror nodes were functioning normally—the first time in at least 10 days—by the same technician in Australia who originally informed Talbot last week that there was a problem.
In an e-mail response from his facility at the Walter and Eliza Hall Institute of Medical Research, in Melbourne, Tony Kyne, the head of information technology services, said that "the replication process has resumed." He added that he received a reply from "a systems manager" at the Hospital for Sick Children, which hosts the site, who said that all mirror nodes have been informed of the renewed service.
Kyne also said he plans to "analyze the database after replication has caught up" to learn if data are properly flowing.
Steve Wallace, the systems manager at the bioinformatics supercomputing center at HSC, in Toronto, confirmed with GenomeWeb on Wednesday morning that the systems were up again.
Asked to explain the nature of the problem, Wallace said: "Glitches happen."
Kyne and Talbot independantly said that the problem originated when the address for the computer that the mirror sites use to download fresh data from the mother site was accidently removed.
There are 11 mirror nodes around the world through which genomics researchers access data from the GDB, which is housed at the HSC. These nodes are copies of the main site in Toronto that scientists can access either to obtain or deliver new data.
Here's how it works: When an update occurs at a mirror node "a piece of computer program is automatically generated" that relays the update to the mother site where it is curated for future use, according to Kyne.
All new data that are generated in any 24-hour period are bundled together into one program, he said. These larger programs are then run against the mother node and are transferred automatically to be run against the mirror sites around the world.
"It is entirely reasonable that human error would have caused what happened, and it was certainly human action… that made it right," Talbot said on Wednesday. "I appreciate the speed with which they [the HSC] addressed it."
Sara Harris in Tokyo and Bernadette Toner in New York contributed to this report.