Using the Neurocurator: Overview
The NeuroCurator has been coded in Python 3 and its user interface was coded with QT. Figure 1 (below) shows the main window of this software.
Figure 1. Main interface of the Curator.
The left panel of this interface provides a view on the Zotero database in use. This database can be setup though the preference dialog which is automatically opened when the program is first ran but can also be brought up through the menu (Edit->Preference).
Figure 2. Preference dialog.
Values shown in this dialog are those proposed by default. They allow to connect the tool to our Zotero database on thalamocortical modeling and to link it to the GIT repository used for persistence and collaborative work. Every modification made to the annotations automatically triggers a GIT commit. A "Push to server" button is provided to manually push committed modifications to the server. Alternatively, when closing the applications after "unpushed" changes have been made, a dialog box automatically pops up asking whether changes should be pushed to the server.
On the Zotero view shown in Figure 1, gray entries are those which have been associated with a PDF; white entries have not been associated yet. Clicking on a white entry will bring up the dialog shown in Figure 3 which allows selecting a PDF file. If no PDF file are available on the local file system, the user can click on the "Follow DOI to the publication website" which opens a new window in the default browser and load the web page associated with this paper (according to the DOI lookup services offered by https://dx.doi.org/). The PDF can be downloaded from there (if the paper is open-access or the user proxy provides a paid access) so that the PDF can be linked with the selected Zotero entry. This button is available only when a DOI is attached to the corresponding publication. Papers with no DOI but with a Pubmed ID can also be added to the database. However, publications no associated with a DOI nor a Pubmed ID cannot, for now, be added.
Figure 3. Dialog displayed when selecting a paper which has not been associated with a PDF in the Zotero view (i.e., selecting a while row).
When a PDF is first associated with a Zotero entry, three files are created in the local repository specified in the preference dialog of Figure 2. These three files have the name of the DOI where "/" symbol has been changed with the "%2F" string. These three files are distinguished only by their extension:
- .pdf : Copy of the original PDF file. These files are automatically save in the GIT repository for convenience.
- .txt : A plain text file created from the original PDF file using a scrapper (i.e., the pdftotext program). These files are used to locate annotations and should be created once and never be modified after that. Any modification that would add or remove characters to these files would compromise localization of the annotations. These files are managed in the GIT repository so that any modification made (by mistake) to these files would be traceable.
- .pcr : PCR stands for "Paper Curation Record". It is a JSON plain text file, which format is described here.
All three files are automatically persisted and versionized in GIT.
Selecting an entry which has been associate with a PDF in the Zotero database (i.e., grey rows) load the existing annotations associated with this publication.
The right panel of Figure 1 shows the paper ID and the list of existing annotations for this paper in the top part. The middle part provide an interface for modifying an existing or adding a new annotation. As specified in the page describing the format of annotations, annotations can be of four types: text, figure, table, equation, and position. The first four should be used when the annotation concern a part of the text, or a specific figure, table, or equation. However, text annotations require the PDF to be in full text. Some older paper are only available as scanned papers. In these case, the fifth kind of annotation ("position") can be used to specified a rectangular bounding bound in a PDF page containing the element to be annotated. However, "text" annotations should be preferred of "position" annotations whenever possible. The lower part contains the controls to add tags or specified modeling parameters associated with the current annotation.
Tags can be added using (see Figure 4) :
a) The combobox: It has been implemented with an autocompletion feature which search (non-case-sensitive) the entered strings in all the Neurolex terms and suggest a list of every Neurolex entities which contain this substring. The list is classified in decreasing order of previous use so that frequently used tags appears first.
b) Suggested tag list: This box contains 10 suggestions of tags. It is populated by computing an index that captures a mixture of both the overall usage of the tags and their usage for this specific paper. Items from this list can be chosen by clicking on them. When clicked, they are moved from the "suggested tag" list to the "annotation tags" list. When clicking on tags in this list while holding SHIFT, the tags will become persistent (indicated by a pale red colouring). That means that they will always figure in the "Suggested tags" list (unless they are already in the "Annotation tags" list). The selection of persistent suggested tags is global (i.e., independent of the paper) and is saved locally only.
c) The Neurolex tags tree: The hierarchy of Neurolex concepts can be browsed using these trees. Those reproduce the architecture that can be browsed in the hierarchies published here. These do not necessarily cover all the Neurolex terms (some of them being available in the tables listed on the same page, which has not been added to this tree widget; those missing terms are nevertheless available through the combobox described in a) ).
Figure 4. Interface for managing tags.
Tags can be removed from the list of selected tags by clicking on them. However, clicking while holding SHIFT will make them persistent for this paper (i.e., they will be entered by default on every new annotation for this paper). If there are already available annotations for this paper, the user will be asked if the persisted tag should be added to these annotations. This preference is saved locally (i.e., it is not saved with the annotation files and it will not persist between different installation of NeuroCurator). Persistent tags are identified by a light red colouring. Clicking on a persistent tag while holding SHIFT will remove its persistence.
The tab for specifying parameters (see Figure 5) is useful to specify explicitly the numerical value of a modeling parameter given in the annotation. These parameter values can then be selected for modeling a network through the MetaModeler.
Figure 5. Interface for specifying values of modeling parameters.