29 July 2007

My NCBI text file toothache

"PubMed, you’ve got a wonderful head on your shoulders. Tell me: whose is it?"
It’s not that I want to harp on about My NeverCanBeInspiring. I have already posted on its virtues, which are amiable though not yet abundant. I am exceedingly fond of the Collections feature, which I use as a kind of super clipboard for merging the results of complex searches. Once a collection is created, I can append citations to it at any time without duplication -- a great convenience when a search takes more than a single sitting to complete.

But all is not well with My NCBI Collections. It's a tooth that’s driving me to extraction. I need relief.

I sense hecklers gathering at their keyboards, grumbling at my nitpicking. But hear me out as I start at the beginning with a run-of-the-mill PubMed search. I’ve done my search, my selections are in the Clipboard, and my Clipboard citations are on the screen. By default they are displayed in reverse chronological order. Something I do frequently is to change the display to Abstract and the sort to journal title order. I then send the results to a text file, which loads into my favourite text editor, PSPad. The display and the sort are preserved. Alles in Ordnung.
Kein Problem.

"And now I live out of all order, out of all compass." (Henry IV, Part 1, 3.3)
Here's the kicker. When I'm working with a search saved in My NCBI Collections, I can't do this. No sorting is preserved when 'sending' to a text file. Through much tweaking and fiddling I have experimented with the different sort options. Yet whatever I do, the downloaded text file is always sorted the same peculiar way: in ascending order by the PMID number. Donner Wetter. This looks like a difficult answer whose question is obvious. I can’t find anything in PubMed's documentation and I can't dig up any discussion of it on the web. Or is this issue so five minutes ago that everyone has thrown up their hands and moved on?

For follow-up: I've had a report from a colleague that sort shortfall (sortfall?) can happen in the regular Clipboard as well. But when I have tried to duplicate the problem on a number of occasions, my Clipboard sorts have been preserved in the downloaded text file. As far as I can tell, it is only in My NeverCanBeImitated Collections that the failure is glaring and persistent.

"Please follow the example of your head and come to the point."

There is a workaround. If you want to get a Collection into a sorted text file, send the citations to Text. For some reason this works:

  1. When working with a search saved in My NCBI Collections, use 'Show' to put all citations into an unbroken sequence (up to 500).
  2. Display and sort the citations as desired.
  3. Select 'Text' from the 'Send to' menu.
  4. Copy the resulting file and paste it into a text editor or word processor.
"So what exactly is on your mind? If you’ll excuse the exaggeration."
My complaint may seem as insignificant as a fly on a puck, but it's more than just a lot of willie wagging. Text file sorting by journal title makes a big difference to my staff. When a search is returned to them with selected articles marked by the patron, they benefit from seeing all the citations from a single journal together. It makes it easier to find those items on the shelves or online, saving time and effort. What’s more, in cases where there are many brief notices on a topic in a single journal issue, the journal title sort keeps them together. If we subscribe to the paper version, we can just lend the whole issue to the patron rather than printing out each item.

"The Tongue is ever turning to the aching tooth."
(Poor Richard's Almanack, 1746)
I realize the topic of today's disgruntlement amounts to little more than a software side note. The sorting problem in My NCBI Collections looks like an oversight on the part of the PubMed developers. Perhaps they will work a fix into a subsequent update, if anyone cares enough. Meanwhile, my tongue has no rest.


Dee said...

Hi Mark, I'm enjoying reading your blog even if I don't get the context of your subtext sometimes. We have had similar issues with Collections where the sort in show and the sort in the collection was different and where the sort was different on a different computer. I have just got to the point where I don't trust it.