PubReMiner Help (FAQ)

How does pubmed PubReMiner work? PubReMiner is a front-end for the popular PubMed literature database at the NCBI. When you submit your query (which can be any query that can be processed by PubMed), PubReMiner will process the result of that query and display its results (in the form of selectable "keywords") in frequency tables, which can be added/excluded from the query to optimize the results.

Why would I need such a tool? PubMed is becoming larger and larger. To obtain a workable amount of references from your queries, often one needs to combine different keywords. But which ones should you use? With this idea in mind, PubReMiner has been developed. The tool allows you to initiate a broad query (which is currently restricted to 7.500 abstracts), after which you can add/exclude words/authors/journals to guide your search. words/authors/journals are displayed in descending order, so you can immediately see which words are used the most in combination with your query. In addition you may also appreciate popular journals in the field of your query, which may help to select a targeted journal for your own work. Furthermore, experts in the field (most actively publishing authors) are becoming visible.

What is this literature mining about? When you are querying a certain subject/gene; the words that are used most frequently often provide a quick way to gather additional information on the subject. if for example we would query the gene "PHOX2B", then we can see immediately in the table words like transcription,neuron,embryology and homeodomain. these terms indicate that PHOX2b is probably a transcription factor, containing a homeodomain which is involved in neurons and probably important during development. Obtaining this insight only took like 3 seconds.

What do those * mean in the words table? * are placed behind words for which different forms have been mentioned in the abstracts. For example if the word "protein" and "proteins" have been observed in the abstracts, then PubReMiner will display "protein *". If you would add this term to your query then both forms are added (protein OR proteins). PubReMiner will try to find the smallest form for every word and groups them on this criterium. This is far from perfect (if the smallest form itself has not been mentioned, then no grouping occurs, but in practice works pretty good). This option is selected by default, but can also be switched off when you don't agree with the grouping.

What is the difference between numart and count? MumArt, displays the number of articles in which the word (or any of the words within a group) has been spotted. Count displays the number of times that a word (or any of the words within a group) has been spotted in total.

How about the selection boxes? The selection boxes within the tables provide the option of "OR" selection. If you select more than 1 box, then in the subsequent query these terms are added as "OR".

Manual adjustment? If for some reason you wish to adapt your query in other ways than PubReMiner allows, then you can adjust the query in any way you like and press "manual". Additionally, if you would like to save you query for use in pubcrawler for example, then you can select the complete query from this box.

Why are some numbers red? If the addition of a word will still result in the same outcome (in other words its addition would not matter, because it is mentioned in every abstract), then its selection is disabled and the number is displayed in red.

In the Mesh terms column some terms contain blue numbers, which I cannot select? The Pubmed engine only allows to search for 1 subterm of a Mesh term (denoted by "/"). However the mesh database does specify more/deeper levels. PubReMiner will desplay those to you, but cannot offer those as a criterium for further reduction. Therefor these are denoted in blue and lack ways to select them. You may use the less defined (2 level) Mesh term for specification though.

I get more results for the Mesh headers than specified. How is this possible? Pubmed uses the hierarchy of MESH to find terms that are encompassed by the selected MESH term. pubreminer only counts terms which are identical. As a consequence, you will get more hits from a MESH term than pubreminer has counted. However, it will still focus the query in the specified direction

What does the blue 'P' mean? The blue 'P' represents a link-out directly to PubMed (which will be opened in a new window). This allows you to query PubMed for all references by an author or Mesh term for example.

I am done with the refinement, where are the abstracts etc? Just press "goto pubmed with query" and off you go.

For whom is pubmed pubreminer developed? Pubreminer was intended for people trying to find biomedical literature on a subject indexed by the NCBI pubmed database (how to find scientific literature). It can help you to find relevant information by suggestions of words that have been used within the abstracts of publications. These keywords often provide a great deal of information on their own. Next to building efficient queries, Pubreminer can also be helpful in other areas: *Selecting a journal for your current work (by scanning the most often used journals of similar research) *Finding experts in a research area (by viewing the authors associated with your query) *Determine the research interest of an author (by viewing the keywords associated with an author)

I need more help or have suggestions? For more help, remarks or suggestions on how to improve the program, emails can be send to jankoster@amc.uva.nl

Jan Koster(AMC) 2004/2007

Version 1.21