Monday, November 19, 2007

Electronic Lab Notebook (ELN) thoughts

My ideal electronic lab notebook would:

  • store everything in real time: I would like to paste in observations, measurements, complete data-processing histories, paper reprints;
  • allow search for: free-text, creation date range, citation date range, citation count
  • allow by-owner stateful highlighting of important things and collapsing of unimportant things: this would facilitate future review without permanently losing observations considered irrelevant at the time. An outline structure would probably be fine, but I would want to be able to collapse or expand any paragraph at any level.
  • allow annotated hyperlinks to prior paragraph-level observations and dates in this notebook and others, for backwards-referencing;
  • automatically count citations (i.e., this "observation was confirmed/rejected" or "protocol was reused" or "data was commented-on" on such-and-such entry) (the two points above combined mean that comments on previous observations, where the comments are actually new ELN entries, remain in-context);
  • I don't see that entries more than two hours old ever need to be edited again, so don't provide an "edit" feature.
  • treat protocols in a special way, somewhat like a source-code control or versioning system. When protocols are mature, it is easy for people to say: I used protocol ABC. However, over the course of years, there could be modifications to protocol ABC that could improve yield, save time, etc. Often, people will continue to say that they used protocol ABC. I would propose that every protocol's "permalink" should contain its version number, so that past links to protocol ABC would point to the historical ABC, not the new-and-improved ABC. Actually, this can be easily handled in the above scheme: just have the complete protocol as an entry; when the protocol is revised, then users of the newer protocol will link to the newer entries. This can be organized with a special "kprotocol" keyword, or a separate "Lab protocols" ELN where protocol developers have write access.
The GUI should facilitate easy linking, perhaps through endnotes for each entry.
When reading, the blog-like "Comment" button would become an "Add new entry to your own ELN" button, with the specified paragraph already referenced. 
When writing, typing a special link string (such as double open braces) should bring up an in-page pop-up allowing search or selection of recently-cited and most-popular ELN citations. While we're at it, this should also allow web search and pasting of URLs or page snapshots.

Discovered that this is an old idea: http://pubs.acs.org/hotartcl/ci/00/jan/inet.html.

No comments: