baderlab / socialnetworkapp Goto Github PK

Query fro the pubmed search bar is:
("Andrews B"[Author] OR "Attisano L"[Author] OR "Audet J"[Author] OR "Bader G"[Author] OR "Blencowe B"[Author] OR "Boone C"[Author] OR "Brown G"[Author] OR "Caudy A"[Author] OR "Chan W"[Author] OR "Emili A"[Author] OR "Fraser A"[Author] OR "Frey B"[Author] OR "Gilbert P"[Author] OR "Greenblatt J"[Author] OR "Hughes T"[Author] OR "Kim P"[Author] OR "van der Kooy D"[Author] OR "Krause H"[Author] OR "Moffat J"[Author] OR "Morris Q"[Author] OR "Morshead C"[Author] OR "Rosebrock A"[Author] OR "Roth F"[Author] OR "Roy P"[Author] OR Ryu WS"[Author] OR "Sefton M"[Author] OR "Shoichet M"[Author] OR "Sidhu S"[Author] OR "Stagljar I"[Author] OR "Taipale M"[Author] OR "Wheeler A"[Author] OR "Yip C"[Author] OR "Zandstra P"[Author] OR "Zhang Z"[Author]) and toronto

Need a more descriptive error. Is the query too long? unrecognized word?

The app should be able to parse reports generated by InCites without requiring any user intervention.

When a network is renamed by right clicking on it and selecting 'Rename Network...' the new name does not get propagated to the network summary panel.

In Incites data when the list of faculty contains people from multiple departments all faculty are annotated with the first department in the file as opposed to the one specified for them.

Use EM annotator to simplify the visualization. --> maybe extract the annotator out in order to use it. (add commands to use the annotator)
Tag clusters with summary tags of the publication represented in the cluster and replace with word clouds (generated by word cloud)

When creating a new network, after selecting the file to use the network name field is entered as N/A instead of the name of the file.

To fix this, usually re-select file and it works on the second time.

Specific for incites data.
Add list of # publication and # times cited to summary panel (and network attributes.

Ideally as a bar chart (but not vital)

At the moment, excluded publications are linked to PubMed with http://www.ncbi.nlm.nih.gov/pubmed/?term=[NAME OF PUBLICATION]. It would be better to perform a search using the PMID like this - http://www.ncbi.nlm.nih.gov/pubmed/?term=[PMID]

Some papers can have more than 1000 authors which produces unmanageable cliques in the network.

Allow the user to set a maximum number that it will compute for. Output the articles excluded based on their thresholds and add an option for the user to select which of those articles they want to included/exclude.

Failed tests:
TestIncites.testIdentifiedFaculty:98 null
TestIncites.testUnidentifiedFaculty:136 null

Tests run: 31, Failures: 2, Errors: 0, Skipped: 0

It is not possible to search for authors whose names contain characters with diaeresis (i.e. Ä or ä)

after building a social network there is information entered in the input panel. If the session is saved and restored this information is missing.

Add the ability to update the static institution to location association file.
For instance, currently Sick kids has no location specification. Right click on author that has an unknown location and add an institution to the set of institution and define which location it should be associated with.
User will then have the option to save this update locally or globally (i.e. save locally to session if it is a change you want only for the given session, save globally when the change is needed for any network created).

Faculty missing are usually associated with a naming issue (i.e. James vs Jim) but faculty (unidentified and identified) don't match up.

Currently pubmed support is through queries to the eutils directly from cyotscape. There is a limit to the length of this query. It would be useful to enable support for pubmed xml files downloaded from the website as the website will allow for much longer queries.

I have already added rudimentary support. If the user specifies the location of a file that end in .xml to the pubmed query bar if it is valid file the app will read it and create a network.

Add this as a proper feature to the interface.

To help make large networks more manageable give the user options on how to filter the edges before the network is created. For instance:
For big papers (with more than X autors):

1. all authors are connected to the last author (instead of all pairwise connections)
2. all authors are connected to central node that is the paper
   Filter out:
3. uncited papers
4. any nodes that are not internal (for the case when you want to see who is collaborating in the same institution)
5. remove connection based on a single paper

The auto-annotator should NOT be hardcoded into the app. Ideally, it should be implemented as its own app and controlled through commands. This makes the whole scheme more modular and easier to maintain.

Also, a tutorial on how to use the auto-annotator viz a viz the social network app should be added to the download page.

Network navigation and any action within cytoscape takes an inordinate amount of time. Check to see if this is a general cytoscape 3.0 issue or if there are optimizations within the code that can fix this.

Sometimes the titles are truncated or have strange characters. This makes the titles difficult to read.

Queries returning a large amount of results may result in NCBI setting limitations on future requests (from a specific IP address). It would be useful if there was a system in place that notified users every time they made a large request and allowed them to back out of it if they liked.

Add a feature that allows users to export the first degree neighbors in a co-publication network. The labels should be the names of the individuals represented by the nodes.

Change Canada in visual style back to red
Associated N/A with its own color (maybe gray) - if there is an issue with mapping the N/A value in cytoscape then when there is no location associated with the default institution set the location to "unknown location"

In physiology network App says there are 51 faculty and 11 weren't identified. In the Network though there are only 36 faculty identified (not the expected 40).
The 4 missing faculty are in the network and in the faculty list but they are not listed in the set of unidentified (and they are not identified as faculty)

Newer publications indexed by PubMed contain the institutional affiliations of all authors who have co-written the publication. It would be nice to include this information in networks generated from this data. When this data is not available, N/A will be used as a stand-in for the institution.

In version 1.0 sent out Sept. 2 the faculty members no longer show up as triangles. Might be related to the recent change that changed the header to Department from faculty.
Suggest we have two columns:

1. Department - department author associated with.
2. Faculty --> yes or no if the author is in the predefined faculty list

A lot of the authors in copublication network built from incites file have self edges (the file lists the same authors multiple times).
When I try to delete the self edges by selecting it and going to Edit-delete selected nodes and edges the self edges get deleted but so do all the edges of the node the self edge is connected to.

Not sure if this is an issue with the App or cytoscape.

Use wordcloud to generate a image representing an author's publications.

I am on Mac OS and for some reason the network summary panel is not being updated properly even though last night, on a Windows machine, it was. From debugging I can see that the correct values are being passed on to the summary panel, but unfortunately they are not being displayed.

If you create a scopus file from merging multiple individual searches of scopus into one main file when the app builds a network it doesn't handle duplicate publication records. It does counts on nodes and edges based on how many hits instead of how many unique hits.

number of copubs should be number of unique co-pubs
number of publications should be number of unique publications.

when a user clicks on a network or when you change between two different social network the Summary window still shows the last network created information.

RIght now users, can only view the excluded publications on a HTML table in the network summary panel and this is not ideal.

Allow the user to enter a subset of people that they wish to see how they collaborate.

Get all papers where there are at least a pair of the given authors on the publication. (ie. given A,B,C search for (A and B) or (A and C) or (B and D) )
highlight the given authors in the resulting network

The app is unable to obtain the relevant help message content through getClass().getResource(). Returns a null every time even though the text files storing the help messages are inside the jar.

Currently to create a sub-network the user selects nodes and edges and uses network->new network->all nodes and edges.
Instead we want a new functionality to:
select nodes and edges we want in the new sub-network
grab the publications represented in that subset
Create a new social network with just those publications (so node citation counts and publication counts will be repflective of the subset and not the original file loaded.)

Sometimes the name of the network is entered at N/A instead of the name of the file so can end up with two faculties "mol gen" and "mol gen " if both are present in the faculty list.

build one network with pubmed xml file
works
try and build a second network from pubmed xml and get the following stack with Loading Netowrk view error:
java.lang.NullPointerException
at org.baderlab.csapps.socialnetwork.tasks.CreateNetworkTask.run(CreateNetworkTask.java:331)
at org.cytoscape.work.internal.task.JDialogTaskManager$TaskRunnable.innerRun(JDialogTaskManager.java:321)
at org.cytoscape.work.internal.task.JDialogTaskManager$TaskRunnable.run(JDialogTaskManager.java:350)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
at java.util.concurrent.FutureTask.run(FutureTask.java:262)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:745)

Every eUtils request must include the tool (name of the app) and email (contact email) parameters. Also, the app must be officially registered (details here -> http://www.ncbi.nlm.nih.gov/books/NBK25497/)

Current each node (author) has the list of publications associated with them. Add an additional field with the count of publications.

Currently you can load a datafile for incites and scopus but not pubmed. Add the option to interface

Instead of only showing only the one institution relating to the location mapping add an additional node attribute called "default institution" that has the preferential institution used for the location mapping.
Change institution to contain a list of all institutions associated with a person.
Document in the user guide the order used when determining the location a set of institutions are related to.

Text mining system that would check various websites e.g. linkedin, google , scholar,pubmed, pubmed central, http://orcid.org/http://www.researcherid.com/, scopus, web of science... and try to identify institution/affiliatioin information for a given author. Start with pubmed as they have added institutional data and see how many holes there are(http://www.nlm.nih.gov/pubs/techbull/so13/brief/so13_author_affiliations.html).

When the user selects any of the google charts they do not launch in the browser.
Looking in the Cytoscape directory the files can not be found. There are just a bunch of framework files that are not open-able.

Add support for Google's new WebGL Globe initiative which is an open platform for geographic data visualization.

More info here: https://github.com/dataarts/webgl-globe

At the moment, the app performs a query to determine the citation for each publication. This is not a good solution since performing too many queries at once is risky - NCBI sets a limit of 3 queries per second.

Add up the counts (publications and citations) for each of the locations but if a publication has 2 ontario authors, 2 canadian authors and 3 international authors then add the one count to each location (instead of adding counts to all locations for each author or the alternative, add counts to the location with the most authors.)

Contact ncbi to find out why?

Add inidvidual search for each pubmed in an xml file to get their counts. Is this too expensive though for time to build?

When looking for a file to load the app defaults to the same main directory instead of remembering the last directory I opened a file from.
Generally input files are found in the same location or similar place and it would be easier to remember previous location.