Hardcore PubMeding (advanced PubMed searching)

From OpenWetWare
Jump to navigationJump to search

PubMed [1] at http://www.pubmed.org is maybe the best search machine for articles in biology. It is run by the NCBI [2], USA, and gives you the possibility to search a gigantic archive of life science literature.

But searching is not easy. Type almost any single term and you will get thousands of hits which you will never be able to browse let alone read. So, to find your preferred needle in the haystack you will need to learn how to wring the best out of PubMed.

Below you find some tips how to become a hardcore PubMeder. Please add your own tricks if you miss them on this page.

MeSH - standardised keywords

The NCBI provides you with standardised search vocabulary. This is advantageous because the right MeSH term (medical subject heading) will find all articles using various synonyms of the term. For example, a brute text search for translation will not return publications using the synonym protein synthesis. The PubMed team annotates all incoming articles with their standardised and hierarchical search terms. You can find out about them by searching the MeSH database [3]. After you have found appropriate terms use them for searching. Type, for example:

protein biosynthesis[mh] stem cells[mh] This returns all publications that have protein biosynthesis (translation) AND stem cells as MeSH terms. PubMed automatically assumes AND as boolean. You can specify OR if required, but connectors have to be capitalised otherwise they are not recognised!
protein biosynthesis[mj] stem cells[mj] Replace the tag [mh] with [mj] to get back only publications that have the term in question as a major (mj) topic. This will significantly reduce the search hits.
  • Note that it is not advisable to put quotes around the search term, like this "stem cells"[mh], since it interferes with the word stemming functions of PubMed. Word stemming is the automatic forwarding of related terms to the main MeSH entry. For example the singular stem cell and the synonym progenitor cell(s) are automatically redirected to the main entry stem cells.
  • Also note that short and long tag names lead to the same results. Instead of [mh] and [mj] you can expand to [MeSH] and [majr] or even [MeSH Terms] and [MeSH Major Topic].
  • Capitalisation is not important for either the search tags or the search terms. stem cells[mh] and Stem Cells[MH] return the same hits.

Besides [mh] and [mj] here are some common PubMed search fields:

[au] author name; e.g. "smith g"[au]
  • searching without quotes will include Smith GX, with X being any initial or none
  • abstain from punctuation - just confuses PubMed; also don't put the initial 1st; Smith, G, Smith G., or G Smith won't work
  • Smith G and author names in the same format are recognised as an author name even without the [au] tag
[ti] article title; e.g. stem cells[ti]
  • note that even without quotes stem cells is considered a phrase
[ta] journal title; e.g. nature[ta]
  • who knows why this field name was chosen; Title of journAl?
[dp] date of publication; e.g. 2006[dp] or 2006/01[dp] for January 2006
  • format is YYYY/MM/DD[dp]; month and day are optional
  • can also be used in different ways: e.g. "last 10 days"[dp] or "last 2 months"[dp] (quotes are required)
  • or 1998:1999[dp] for all publications from '98 to '99

See this PubMed help page for a complete list.

PubMed is trying hard to guess what you were really looking for. To find out how the search engine interpreted what you typed, click on the "Details" tab after the search has been run.

My NCBI - personalising PubMed

PubMed offers you the possibility to save your searches for reuse and automatic email alerts. You need to register for this option. My NCBI [4] also allows you to set up filters [5], define document delivery options, and set user preferences. You will find the link on the left of PubMed main search page [6]

Related Articles function

PubMed includes a very useful function to suggest related articles to the one(s) you just retrieved by a search. The feature relates articles by comparing their words [7]. You will find them on the right of the abstract page. Often this will help you to find article not spotted by your search but of interest to you.


Automatic search string interpretation

PubMed doesn't just search 1:1 for what you type. Instead it processes your search string and tries to return the best results. Usually, this helps, esp. if you are not using field identifiers as described above. However, sometimes in can create problems. Try searching for the author "Black" without field names by just typing Black into the search box. Search and then click the Details tab. This tells you how PubMed interpreted your search string. You will see the following:

"african continental ancestry group"[MeSH Terms] OR 
("african americans"[TIAB] NOT Medline[SB]) OR 
"african americans"[MeSH Terms] 
OR Black[Text Word]

Not exactly what you wanted probably. Black is a synonym for the MeSH term African American. It actually doesn't return any hits for Black as an author. You can remedy the situation either by giving initials Black S or by appending the [au] field name.

Copying and pasting to PubMed

You just spotted an interesting paper in the reference list of a paper. So, you set about to copy and paste that reference to PubMed to read the abstract and copy it into your reference manager.

This method often doesn't work because PubMed is quite picky about formating. The easy solution is to use Google Scholar instead which is much more flexible. From there you can go back to PubMed if you need the PMID for example.

To illustrate the problem, here are 2 examples:

  1. try searching for "Venter C. et al. Science" - no results since PubMed chokes on the et al.; if you copy from journals that abbreviate their author lists using "et al."; again the Details tab is your best friend to find out why this didn't work: "Venter C.[Author] AND et al.[Author] AND ("science"[MeSH Terms] OR Science[Text Word])"
  2. try "Laemmli, U.K. T4" - 1 result, but not from Laemmli; the Details show: "Laemmli[All Fields] AND U.K.[All Fields] AND T4[All Fields]"; Here the comma is the main problem due to which initials and last names are not recognised as belonging together. Remove it and you'll get the right results. Punctuation is generally a bad idea when searching PubMed. It's usually best just to remove any.

Further reading