Google Trick for Science Search
While specialised search engines will always have their place, I find myself drifting more and more toward google for the thorough, 'all document' search. Here is a simple example - I want to know if SCN2A (the gene encoding the Nav1.1 channel) has been expressed in HeLa cells. My first destination is PubMed:
- SCN1A HeLa → 1 result, a 'false' positive
Next, I'll try google:
- SCN1A HeLa → 6070 results
Far too many to search through. Most of the results appear to be mentions of one or both words on the same web page.
If I think about what I am actually search for it is: "We expressed SCN1A in HeLa cells…". I could search for that exact expression (0 results), but I know that there are a hundred different ways of saying the same thing. I would like to search for all those web pages where SCN1A is in the same sentence as HeLa cells. There is at least one commercial solution out there - the excellent DevonAgent, which I suggest is worth a look. But a similar, though limited, functionality is available in Google.
- "SCN1A * HeLa" → 1 result, a 'false' positive
(Enter with the quotation marks). This produces all the webpages with SCN2A preceding HeLa in the same sentence; of course HeLa may precede SCN2A, so I also need to enter the search phrase:
- "HeLa * SCN1A" → 0 results
This gives me some confidence that SCN1A has probably not been expressed in HeLa cells; at least not documented to have been expressed. I realise that this method may result in some misses - when HeLa and SCN1A are split by a single period for instance. But it is a far better method than the standard google search and has many uses beyond science.
I hope for the day when one can supply a search modification in the google search bar along the lines of 'search for these two words, and only return those webpages where they are within [x] words of each other'.
