eToc Replacement Therapy with Sente

Keeping track of the current literature is a Sisyphean task, and my approach to it continues to evolve. Whereas once I thumbed the contents pages of hard copies of journals as they were delivered, my main mode of contact is now electronic, and yet keeping up remains a struggle. In this article, I briefly describe a number of my now discarded strategies (roughly in chronological order) before ending up with my current favorite, using a fantastic, Mac-only program called Sente to manage the current literature.

Table of contents

The trouble with eTocs

Emailed Tables of Contents—eTocs—were once a marvelous invention. You could sign up for them at most journals (as an example, here's the sign-up page at PNAS), and each journal would automatically send you an email when a new set of articles became available. However, it didn't take long—probably after signing up for my tenth journal—for the email traffic to become a burden. Sure, I set up filters to automatically move the eToc messages out of my inbox and into a dedicated mailbox—this freed me from having to sort the messages and getting distracted by a flood of new articles while I was looking for email from my department chairman—but it didn't really solve two other problems:

  1. Keeping Track of My Reading Trail. I wanted to document that I'd actually looked at the journal issue, and I wanted something better than either deleting the mail message once I'd looked at it (as sensible as that seems), or moving it to yet another mail box. I even went to the length of building a FileMaker database in which I could (relatively) easily record which journal issues I'd looked at, but even this was a burden.
  2. Managing Citations. I wanted an easy method for collecting citations to articles of interest in a dedicated citation manager such as EndNote, or now, Sente. On this point, eTocs are a complete failure. At best, you have to look up the article in PubMed, but in some cases, the articles in the eTocs aren't yet available in PubMed.

Oh, and the other problem with eTocs is that they are "Pushy" — anyone remember the Push technology bubble, where people (not me!) had screen savers with stock tickers and CNN crawlers. I always thought it was a stupid idea, and an ecological catastrophe; thankfully, the fad passed quickly. eTocs aren't quite as bad, but the idea is similar.

Using a tabbed browser

If you're not familiar with tabbed internet browsing—available in Safari (where it must be turned on in the preference panel), FireFox, Netscape, OmniWeb, and other browsers (but not current versions of Internet Explorer)—you're probably making your life harder. For a while now, I've collected URLs for related sets of journals in a bookmark folders, and then I open them all up at once in a set of tabs. I would then scan the Tables of Contents, and peruse the articles online, before downloading the PDFs of those that I wanted to keep on my computer. This was a more pleasant experience than reading emailed eTocs, but it didn't solve the "Keeping Track of My Reading Trail" that I described above. In addition, if I wanted to keep track of the citation in EndNote or Sente, I had to download each citation individually. When you're a PDF pack rat like I am, this gets old, quickly

Sente to the rescue

Sente is a fantastic new reference manager application and has completely replaced EndNote for nearly all of my reference managing and searching needs. In the near future, I plan to write up a more complete description of why I like it so much and how I use it, but for now, I will restrict my comments here to a description of how Sente can be used to keep track of the current literature.

Automated updating of searches is the key

One of the best features of Sente is that all PubMed searches that you perform using it are kept in a database; furthermore, these searches are automatically updated every day. So, if you set up a search to look for all articles about HIV AND CTL, Sente will automatically run this search once a day, and will update its database with articles that have been added since the last search; this is called a Current Awareness Search, and in Sente, all saved searches are Current Awareness Searches.

Normally, I use Current Awareness Searches for topics, such as in the HIV AND CTL example above. However, it recently occured to me (duh!) that I could use a Current Awareness Search strategy to keep track of articles as they are added to PubMed—using journal names as search terms. Here's how I set up the search dialog box to do it in Sente:

In this example, I have set up a search that finds articles in The Journal of Clinical Investigation. In order to limit the number of articles coming in when the search is first run, I toggled the dialog box to "Show Advanced Options", and I set it up to retrieve only articles published after 07/15/05. On the day that I set this up, it retrieved 56 references, as indicated in the figure below.

Note the Golden Bullets (), which are Sente's visual cue that I have not yet opened that Sente record. When I first set up a search, I scan the list of all of the articles, then I select them all and activate the "Set Selected to Viewed" command from the Edit menu, which removes all of the Golden Bullets. On a future date, when Sente automatically adds new articles, only the new articles will be marked with the Golden Bullets (as will the whole collection, in the pane at the left in the figure above). The easist thing to do, at that point, is to sort the collection by they "viewed status", simply by clicking on the Golden Bullet at the top of the references pane (clicking it toggles the sort order, so you can play around with this to get references that you want at the top).

You'll note a few other things in the figure above:

  1. I have organized the searches in a hierarchical order, taking advantage of one of the most powerful features of Sente. The name of the collection at the top level of the hierarchy is 003 eTocs; this type of collection is called a User Collection, and is denoted with the icon. The reason for prepending "003" is because Sente sorts collections alphabetically, and this forces the collection to the top of the sort list.
  2. Next in the hiearchy, you'll see collections labeled Biochemistry; Cell, PNAS, Science; Clinical Journals; Immunology Journals; Nature Journals; Nature Reviews (the separation from Nature Journals is arbitrary); PLoS; and Virology. All of these are somewhat arbitrary, but I like to impose some order. These are also User Collections.
  3. You'll note that the J Clin Invest collection is a standard "Search Collection" denoted by the icon , but that EMBO J is a filter collection (denoted by the icon) . This is because when I first set up the Biochemistry search, it was set up to look not just for articles in the journal Biochemistry, but EMBO J, J Biol Chem, J Mol Biol, etc. as well. This search collects all of the articles in one search collection, and I then set up filters to deconvolute them, as in the following figure:

    4. I'm not sure that one method of doing this is preferable to another—setting up searches for individual journals vs. setting up searches for collections of journals, and then deconvoluting them (if desired) with Filters—it just shows that there is more than one way to skin a cat. I started this with the combined searches, but now I tend to set up searches for individual journals and then move them to an appropriate parent collection.

    Update (9/6/2005)
    I just discovered the reason why it is NOT a good idea to lump all of the searches together and then set up filters to deconvolute them. There is a natural temptation to select the filter collection, and to delete the references you don't want. However, since this is a filter, this doesn't work. I now suggest that you set up searches for only one journal at a time, and then place those searches inside an appropriately named parent folder.

How do I keep all of the searches organized?

To review, the left-hand pane of the main Sente window contains a list of four distinct types of collections:

  1. Search collections (). These are created when searching an on-line database, such as PubMed.
  2. Filter collections (). These are a subset of any other type of collection. They require a definition, as above.
  3. User collections (). These are built manually by dragging a reference from any other type of collection, or by importing from another file. They can contain references, or other child collections, as above.
  4. Bibliography collections. These are like user collections, but Sente builds a formatted bibliography for them, using pre-programmed bibliography formats; a limited set of these come with Sente, but it's not too difficult to build your own.

The full collection list is sorted alphabetically, and it is not possible to manually change the order of the collections, or even to manually group together like sets of collections, such as Bibliography Collections, or Search Collections of a distinct sort. However, it is possible to set up a hierarchical organization of collections, such that one collection might be a child of another; in this way, a set of collections can easily be grouped as "siblings". In the example above, I have created a parent User Collection called "003 eTocs", and I have made several Search Collections children of this (e.g. Biochemistry; "Cell, PNAS, Science"; Clinical Journals; Immunology Journals; Nature Journals; Nature Reviews; PLoS; Virology). The name "003 eTocs" was used to force this whole collection, children and all, to the top of the Sente collection list.

How did I set up the names of the searches?

When you set up a new search, for example, looking for articles published in The Journal of Biological Chemistry, Biochemistry, J Mol Biol, or EMBO J, Sente will give the search a default name based upon the search terms, such as "J Biol Chem, ...". There are three ways of calling up the "Rename" command that will let you set the name of this collection to something like "999 Journals - Biochemistry". Once you have selected the collection, you can then:

  1. Select the "Rename" command from the File menu.
  2. Select the "Rename" command by typing command-R.
  3. Control-click (or right-click) on the collection name, and select the "Rename" command from the contextual menu.

Once you've done this, you simply type in the new name.

Why bother collecting all of these references when they are mostly irrelevant?

That's an excellent question, with a number of answers. The most important answer is that this is the most efficient way I have come up with to do all of the following: (1) scan the new literature; (2) download the articles that I want; (3) build a database of articles that I have on my computer; and (4) allow me to use that database to get to the articles quickly. Here's how I do it.

  1. Select an article I want to read from a Sente search collection.
  2. Select the PubMed LinkOut command from the "View: External Links" menu. This opens your default Internet Browser to the PubMed Linkout page corresponding to the article.
  3. Click on the "Full Text" link in the PubMed window. This takes you to the journal page for the article. Unfortunately, it's a little tricky to go there directly there from within Sente, but it can be done.
  4. Click on the "PDF" link to download the article.
  5. Rename the PDF file and save it in an appropriate directory on your hard drive.
  6. Drag the icon for the file from the directory on your hard drive onto the record in Sente. This tells Sente where the files is on your disk. This takes a little bit of gymnastics to get the Finder folder and the Sente window available at the same time, but you'll quickly master it. Once you have done this, instead of the record being marked with a Golden Bullet, it will be marked with a Blue Bullet, indicating a link to a local file. Watch the magic when you double click on the Blue Bullet. You'll be amazed.
  7. Drag the record from the search collection into your Master User Collection, or any other collection.

This Seven Step procedure may sound complicated, but it beats downloading the article, and then setting up a specialized PubMed search to retrieve a Sente record corresponding to that article.

There are three other answers to this question. First, the Sente searches and record retrievals are very fast. I wouldn't do this using EndNote, because it's such a snail at record retrieval, but this is not an issue at all with Sente. Second, I don't really care about the space it takes up on the hard disk, and Sente does all of the work for me of managing the database. Again, contrast this to EndNote, in which you have to manually manage individual files corresponding to each of the searches (and it's a bit moot, since EndNote doesn't automatically update searches anyway). Finally, the whole procedure gives me a record of issues that I've looked at, and I can throw away the FileMaker databases that I built, and that were a pain from the data-entry point of view, anyway.

Can I delete articles from searches if I no longer need them?

Sente will certainly allow you to do this, and it won't re-retrieve the deleted articles, but apparently, PubMed sometimes updates records, and I think these updates might get picked up in Sente.

What happens with Advanced Online Publications?

PubMed now includes links to Advanced Online Publications (AOPs). Whether this is an advance or not is a matter of debate. There are two potential problems:

  1. Just yesterday, PubMed had records up to a new issue of Immunity that the Immunity site didn't yet have. This is a small problem, and just requires a little patience.
  2. More importantly, from the point of view of building a list of citations that can be used to build a bibliography, records of the final publication are much more appropriate than records of AOPs. I don't yet know how Sente will handle this.

Note that from the point of view of storing PDFs of AOPs on your computer and creating links to them in Sente, there is no problem at all.