Unlike the Agatha Christie mystery referred to in the title of this piece, there are some mysteries that seem to defy a single answer. One of those mysteries is how most scientists manage to keep up-to-date given the growing avalanche of developments in their fields. The answers I get are wide and varied, but I thought I would share a couple variations on my workflows, and perhaps start a conversation at the same time.
The first step in the process is to find articles. I use two different strategies for this. The first I call “Surveillance”. I want to have a broad understanding of what’s happening in several different (but related) areas. In my case, although I’m particularly interested in pancreatic cancer, its disease processes and target space, I’m also interested in other cancers with an overlapping target space, and the rationale behind new treatments.
I start by selecting a few genes from pancreatic cancer, and search for papers that do not include “pancreatic cancer”. Since KRAS plays a vital role in the etiology of pancreatic cancer, I add the term “etiology” to see what other diseases it may be involved in. My results then include diseases like colorectal cancer, endometrial cancer, and non-small cell lung cancer, etc. I usually narrow this further by only including articles about clinical trials because I want to see if drugs used to treat other indications, might also be used in pancreatic cancer. After quickly reviewing the abstracts, of some promising candidates, I take a close look at the journals where the articles are published, and look at the Journal TOCs site to see what the current Table Of Contents for each journal. I also check the AOP (Advance Of Print) contents, to get a heads-up
I then add the RSS feeds found in the Journal TOCs site, to my Feedly account. This lets me aggregate and monitor the feeds.
Ordinarily, you might stop with PubMed; however, I’m often interested in developments that make it to the news, or about other institutions that may be investigating pancreatic cancer. You can go to the Google News web site and create these searches. You can create an alert for the search and the results will be delivered via RSS, or email. I usually choose RSS, because I want the results to be added to my Feedly account and to be grouped with related feeds. This lets me group KRAS-related journal searches, with KRAS-related news articles.
I also use a more “Directed Search” strategy. I define targeted searches in PubMed to which I subscribe to via email. These searches are focused on specific disease processes — “pancreatic cancer AND perineural invasion” or “pancreatic cancer AND desmoplasia”. They allow me to identify particular targets that may play a critical role in a specific disease process.
Pocketing the results
From these different search strategies, you end up with lists of papers to be reviewed. In the old days, I might print some of the more interesting papers, scribble the PMID on the title page, and read them later when I had a free moment. The problem was always that the printed article was never at hand when that free moment arrived.
With Pocket, though it never really matters. I always have a phone, tablet or computer at hand, and any article that I “Pocket” is available for offline reading. This feature has come in handy on numerous flights, or simply waiting for a meeting to start.
Pocket also lets you tag the papers that you save. This makes it easy to group papers together by disease process, or pathway.
In addition to Pocket, I’ve found myself using Paperpile more frequently. In particular, Paperpile makes easy to share the paper with colleagues and to cite it. Usually what goes into Paperpile, are those papers that are directly related to a project that I’m working on. Paperpile has a browser extension that adds a little “add to Paperpile” button next to PubMed search results, which makes it pretty easy to add those articles to Paperpile.
I’ve also tried Evernote, and Google Keep, but I invariably return to Pocket and Paperpile. The only challenge to using these generic tools, is that they aren’t aware of semantic concepts. This means that I can’t cluster papers by MeSH term, or simply use a MeSH term from a pocketed article to do a “More like this” search.