Figure 1.
This sequence illustrates how posts develop as a process is recorded or minor mistakes are corrected.
The LabTrove system archives all versions of the post and they are accessible from the final version of the post. All the URLs resolve to a open science laboratory notebook and should be visible. (A) The first version of a post describing an analysis procedure (http://biolab.isis.rl.ac.uk/camerons_labblog/11928/Some_analysis_of_reflection_data_from_Crisp_Experiment.html?revision=11928 Accessed 6 March 2013). (B) The record of revisions, showing the first five revisions of the post (http://biolab.isis.rl.ac.uk/camerons_labblog/11928/Some_analysis_of_reflection_data_from_Crisp_Experiment.html?revisions, Accessed 6 March 2013). (C) The final version of the post including further analysis and images (http://biolab.isis.rl.ac.uk/camerons_labblog/11928/Some_analysis_of_reflection_data_from_Crisp_Experiment.html?revision=11938, Accessed 6 March 2013). Copies of these web pages are provided as part of the supplementary material as well as the links to the Labtrove instance at ISIS.
Figure 2.
Extract from a laboratory notebook recorded as a blog, presented in a standard format, and showing three individual posts.
The top two posts represent samples whereas the bottom post is a procedure. This view is accessible at http://biolab.isis.rl.ac.uk/camerons_labblog/month/1280617200, Accessed 6 March 2013.
Figure 3.
Tools used when adding and editing posts.
(A) The editing view for creating text content, with a toolbar for markup, which includes an icon for linking to another post. (B) Clicking the link icon pops up a window from which an existing post can be selected to incorporate a link into the current post. (C) Data files can be added only after a post has been submitted. A popup window enables the selection and naming of the appropriate files, which are then attached to the post. (D) Thumbnails of the files can also be added into the text of the post.
Figure 4.
Diagram illustrating the alternative schemes for determining which research objects merit their own posts.
Strictly, scheme (A) is more correct, as every individual process or reaction receives its own post and therefore has its own URI. This approach has the advantage of creating a discrete endpoint for every relationship between inputs, processes, and outputs. However, this scheme has the potential to create an unsustainable burden on the user and to render the records difficult to read as a laboratory notebook. In scheme (B), which we usually adopt, objects (in this case, samples and data) have their own posts but repeated processes are collected together and the relationship between specific inputs and outputs is recorded implicitly in a table. Posts that use this approach are more human-readable, but if necessary specific values can still be extracted programmatically into structured data.
Figure 5.
Extraction of unstructured and partially structured data from LabTrove posts.
Similar posts, e.g. those that refer to a single type of material can be readily aggregated and parsed based on local knowledge of the structure. This information can then be tabulated or used to populate a relational database. It is also possible to extract parameters, such as the sequence of an oligonucleotide, do external analysis, and pass this information back into the post. See Table 1 for an example of structured data obtained from posts that contain tables and the details of the methodology.
Table 1.
Extraction of structured data from partially structured LabTrove posts.
Figure 6.
Extracts illustrating the design and use of templates.
(A) A template post for a data collection type experiment in which a sample is measured with specific instrument parameters. The template includes placeholders for links to other posts in the ‘Sample’ column - [[Material:Solution]] – and in the ‘Data’ column. Note that the data itself would be created after the current template was used, so the post identifier data would be inserted during a subsequent edit. (B) The rendered template showing drop-down menus for selecting sample posts and text boxes for entering values. (C) A template for an experimental laboratory procedure involving a range of input samples and a single output product. (D) The rendered template being populated.
Figure 7.
Visualization of posts as a network of resources.
The full set of posts from a single e-Notebook was obtained as an XML document and parsed [30] into Geph (http://gephi.org/) graph format [31]. A subset of posts from a specific time frame, during which a number of protein purifications were being carried out, was selected and automatically grouped by sub-network modularity. The modularity analysis successfully differentiates the posts in three separate but parallel experiments carried out at the same time. The inset shows the same graph colour-coded by post type: pink nodes are procedures, yellow nodes are physical samples, and the red node is a dataset.
Figure 8.
Example of a LabTrove TimeLine.
Each item on the TimeLine is a hyperlink to the LabTrove entry.